Draft version December 3, 2021Typeset using LATEX twocolumn style in AASTeX63

Galactic Winds across the Gas-Rich Merger Sequence.I. Highly Ionized N V and O VI Outflows in the QUEST Quasars∗

Sylvain Veilleux,1, 2 David S. N. Rupke,3 Weizhe Liu,1 Anthony To,3, 4 Margaret Trippe,1, 5, 6

Todd M. Tripp,7 Fred Hamann,8 Reinhard Genzel,9 Dieter Lutz,9 Roberto Maiolino,10, 11 Hagai Netzer,12

Kenneth R. Sembach,13 Eckhard Sturm,9 Linda Tacconi,9 and Stacy H. Teng1, 14

1Department of Astronomy, University of Maryland, College Park, MD 20742, USA2Joint Space-Science Institute, University of Maryland, College Park, MD 20742, USA

3Department of Physics, Rhodes College, Memphis, TN 38112, USA4Department of Physics, University of Hawaii, Honolulu, HI 96822, USA

5Johns Hopkins University Applied Physics Laboratory, Laurel, MD 20723 USA6Lincoln Laboratory, Massachusetts Institute of Technology, Lexington, MA 02421-6426 USA

7Department of Astronomy, University of Massachussetts, Amherst, MA 01003, USA8Department of Physics and Astronomy, University of California, Riverside, CA 92507, USA

9Max-Planck-Institut für Extraterrestrische Physik, Giessenbachstrasse 1, 85748 Garching, Germany10Cavendish Laboratory, University of Cambridge, 19 J.J. Thomson Avenue, Cambridge CB3 0HE, United Kingdom11Kavli Institute for Cosmology, University of Cambridge, Madingley Road, Cambridge CB3 0HA, United Kingdom

12School of Physics and Astronomy, Tel-Aviv University, Tel Aviv 69978, Israel13Space Telescope Science Institute, Baltimore, MD 21218, USA14Institute for Defense Analyses, Alexandria, MD 22311, USA

ABSTRACTThis program is part of QUEST (Quasar/ULIRG Evolutionary Study) and seeks to examine the

gaseous environments of z . 0.3 quasars and ULIRGs as a function of host galaxy properties andage across the merger sequence from ULIRGs to quasars. This first paper in the series focuses on33 quasars from the QUEST sample and on the kinematics of the highly ionized gas phase traced bythe N V λλ1238,1243 and O VI λλ1032,1038 absorption lines in high-quality Hubble Space Telescope(HST) Cosmic Origins Spectrograph (COS) data. N V and O VI outflows are present in about 60% ofthe QUEST quasars and span a broad range of properties, both in terms of equivalent widths (from 20mÅ to 25 Å) and kinematics (outflow velocities from a few × 100 km s−1 up to ∼10,000 km s−1). Therate of incidence and equivalent widths of the highly ionized outflows are higher among X-ray weakor absorbed sources. The weighted outflow velocity dispersions are highest among the X-ray weakestsources. No significant trends are found between the weighted outflow velocities and the properties ofthe quasars and host galaxies although this may be due to the limited dynamic range of propertiesof the current sample. These results will be re-examined in an upcoming paper where the sampleis expanded to include the QUEST ULIRGs. Finally, a lower limit of ∼0.1% on the ratio of time-averaged kinetic power to bolometric luminosity is estimated in the 2−4 objects with blueshifted P Vλλ1117,1128 absorption features.

Keywords: galaxies: evolution − ISM: jets and outflows − quasars: absorption lines − quasars: general

Corresponding author: Sylvain Veilleuxveilleux@astro.umd.edu

∗ Based on observations made with the NASA/ESA Hubble SpaceTelescope, obtained from the data archive at the Space TelescopeScience Institute. STScI is operated by the Association of Uni-versities for Research in Astronomy, Inc. under NASA contractNAS 5-26555.

1. INTRODUCTION

Large multiwavelength surveys of the local and dis-tant universe have shown that major mergers of gas-

arX

iv:2

112.

0078

9v1

[as

tro-

ph.G

A]

1 D

ec 2

021

2 Veilleux et al.

rich galaxies1 may trigger spectacular bursts of star for-mation, accompanied with quasar-like episodes of rapidgrowth of the supermassive black holes (SMBHs), andresult in merger remnants that follow tight SMBH-hostscaling relations and resemble today’s quiescent early-type galaxies (e.g. Sanders et al. 1988; Hickox & Alexan-der 2018). Modern simulations of galaxy formation andevolution (e.g. Nelson et al. 2019; Oppenheimer et al.2020; Nelson et al. 2021) largely reproduce these obser-vations. However, the root cause of the fast “quenching”of the star formation activity in the merger remnantsdepends on the detailed, sub-grid scale, implementationof how the mass, momentum, and energy from stellarwinds, supernova explosions, and SMBH-related pro-cesses are injected into, and interact with, the interstel-lar medium (ISM) and circumgalactic medium (CGM)of the host galaxies. Over the past several years, nearbygas-rich galaxy mergers have emerged as excellent labo-ratories to study in detail these stellar and quasar feed-back processes (see Veilleux et al. 2020, for a recent re-view). These objects are the focus of the present study.Locally, major gas-rich galaxy mergers often co-

incide with obscured ultraluminous infrared galaxies(ULIRGs). As these systems evolve, the obscuring gasand dust, funneled to the center by the dissipative col-lapse and tidal forces during the merger, are either trans-formed into stars or expelled out of the nucleus by pow-erful winds driven by the central quasar and starburst,giving rise to dusty quasars and finally to completelyexposed quasars. Galactic-scale winds are ubiquitous inlocal ULIRGs and dusty quasars (e.g. Sturm et al. 2011;Veilleux et al. 2013a; Cicone et al. 2014; Rupke et al.2017; Veilleux et al. 2017; Fluetsch et al. 2019, 2020;Lutz et al. 2020; Veilleux et al. 2020, and referencestherein). The outflows detected in ULIRGs extend overa large range of distances from the central energy source,seamlessly blending with the circumgalactic medium at>10 kpc (Veilleux et al. 2020, and references therein).In these objects, the outflow masses and energetics are

often dominated by the outer (&kpc) cool dusty molec-ular or neutral atomic gas phase, but the driving mech-anism is best probed by examining the inner (. sub-kpc) ionized phase. ULIRG F11119+3257 is the firstand still the best case among local ULIRGs where a fast(>0.1 c), highly ionized (Fe XXV/XXVI at ∼7 keV),accretion-disk scale (<1 pc) quasar wind appears to bedriving a massive (> 100 M yr−1), large-scale (1−10+kpc) molecular and neutral-gas outflow (Tombesi et al.

1 In this paper, we define major mergers as those involving galaxieswith . 4:1 stellar mass ratios.

2015, 2017; Veilleux et al. 2017). Unfortunately, thesearch for hot winds in a statistically significant sampleof ULIRGs is not feasible at present since most ULIRGsare too faint at ∼7 keV for current X-ray observatories.This is where the excellent far-ultraviolet (FUV)

spectroscopic sensitivity of the Hubble Space Telescope(HST) becomes handy. The FUV band is rich in spec-troscopic diagnostics of the neutral, low-ionization, andhigh-ionization gas phases (Haislmaier et al. 2021); thusHST can probe all three phases at once. So far, onlyabout a dozen ULIRGs and IR-bright quasars havebeen studied with HST, but the results have beenpromising. Prominent, blueshifted Lyα emission outto −1000 km s−1 has been detected in half of theseULIRGs. Blueshifted absorption features from high-ionization species like N V and O VI (77 and 114 eVare needed to produce N4+ and O5+ ions, respectively)and/or low-ionization species like Si II, Si III, Fe II,N II, and Ar I have provided additional unambiguoussignatures of outflows in a few of these objects. Martinet al. (2015) have argued that the FUV-detected out-flows represent clumps of gas condensing out of a fast,hot wind generated by the central starburst (Thomp-son et al. 2016). This picture is also consistent with theblast-wave model for quasar feedback. In this model, afast, hot wind shocks the surrounding ISM, which theneventually cools to reform the molecular gas after havingacquired a significant fraction of the initial kinetic en-ergy of the hot wind (e.g. Weymann et al. 1985; Zubovas& King 2012, 2014; Faucher-Giguère & Quataert 2012;Nims et al. 2015; Richings & Faucher-Giguère 2018a,b;Richings et al. 2021; Girichidis et al. 2021). An alterna-tive explanation is radiative acceleration (e.g. Ishibashiet al. 2018, 2021), which may dominate the dynamicsof outflows on a wide variety of scales (e.g. Stern et al.2016; Revalski et al. 2018; Somalwar et al. 2020).So far, the published data set on ULIRGs and

IR-bright quasars is too small to draw strong con-clusions about the properties of the FUV-detectedwinds. There is tantalizing evidence that UV-detected AGN/starburst-driven winds are present inmost ULIRGs, but the sample is very incomplete, partic-ularly among ULIRGs with AGN and matched quasars.A more diverse sample of ULIRGs and quasars is neededto study the gaseous environments of nearby quasars andULIRGs as a function of host properties and age acrossthe merger sequence from ULIRGs to quasars. This is-sue is addressed in the present study.In this first paper, we focus our efforts on studying

the highly ionized gas, traced by N V λλ1238, 1243 andO VI λλ1032, 1038, in a sample of 33 local quasars,while Paper II (Liu et al. 2021, in prep.) will present

Highly Ionized N V and O VI Outflows in the QUEST Quasars 3

the results on our sample of ULIRGs with AGN andcompare them with those on the quasars. As statedin Hamann et al. (2019b), the quasars in the presentsample are valuable for outflow studies in and of them-selves because: 1) they fill a largely-unexplored nichebetween luminous quasars with strong broad absorptionlines (BALs) with outflow velocities of up to 0.1 − 0.2c and low-luminosity Seyfert 1 galaxies with exclusivelynarrower outflow lines, 2) their low redshift minimizescontamination by the Lyα forest, and 3) the outflowlines are relatively narrow so blending is less severe. In-deed, as we discuss below, the detected outflows oftenare “mini-BALs” instead of BALs because their velocitywidths lie below or near the threshold of 2000 km s−1

used for BALs (Weymann et al. 1981, 1991; Hamann &Sabra 2004; Gibson et al. 2009b).Our quasar sample is discussed in Section 2. The ex-

tensive set of ancillary data on these quasars is summa-rized in Section 3. The HST spectra used for this studyare described in Section 4 and the methods applied toanalyze these data are detailed in Section 5. The resultsfrom this analysis are presented in Section 6, and dis-cussed in more detail in Section 7. Section 8 provides asummary of the main results from this paper.

2. QUASAR SAMPLE

The quasars in our sample are selected using fourcriteria: (1) They must be part of the QUEST(Quasar/ULIRG Evolutionary Study) sample of local(z . 0.3) ULIRGs and quasars. The QUEST samplehas already been described in detail in Veilleux et al.(2009a,b) and references therein. All 33 objects in thepresent sample are Palomar-Green (PG) quasars fromthe Bright Quasar Sample (Schmidt & Green 1983), ex-cept Mrk 231, the nearest quasar known, whose UVspectrum has already been analyzed by Veilleux et al.(2013b, 2016) and will not be discussed here any further.As part of the QUEST sample, the quasars are carefullymatched in terms of redshifts, bolometric luminosities,and host galaxy masses with the QUEST ULIRGs of Pa-per II. (2) Their bolometric luminosity must be quasar-like, & 1045 ergs s−1, and dominated by the quasarrather than the starburst based on the Spitzer data (seecriterion #3 below) or, equivalently, have 25-to-60 µmflux ratios f25/f60 & 0.15 (Veilleux et al. 2009a). Thiscriterion also automatically selects UV-detected late-stage mergers or non-mergers (Veilleux et al. 2009b). (3)A strong preference is also given to the QUEST quasarswith Spitzer mid-infrared spectra to provide valuable in-formation on the AGN contribution to the bolometric lu-minosities of these objects. (4) High-quality COS spec-tra covering systemic N V and/or O VI must exist for

each object in the sample. Only COS data are consid-ered to ease comparisons between spectra and avoid pos-sible systematic errors associated with comparing datasets from different instruments. As described in Sec-tion 4, both our own and archival data are used for thisstudy.These criteria result in a sample of 33 objects. Table

1 lists the key properties of the quasars in our sam-ple, many of which are derived from our extensive setof ancillary data on these objects, discussed in Section3. As shown in Figure 1, these quasars cover the low-redshift and low bolometric luminosity ends of the PGquasar sample. They are well matched in redshift withthe QUEST ULIRGs which will be studied in Paper II(Liu et al. 2021, in prep), and are representative of theentire PG quasars sample in terms of infrared excess(defined here as the infrared-to-bolometric luminosityratio, LIR/LBOL) and FIR brightness (L60 µm/L15 µm

from Netzer et al. 2007).

3. ANCILLARY DATA

An extensive set of spectroscopic and photometricdata exist on all of the objects in the sample. SloanDigital Sky Survey (SDSS) optical spectra are availablefor all of them. High-quality optical spectra also existin Boroson & Green (1992), and Krug (2013) presentsspectra centered on Na I D λλ5890, 5896. As mentionedin Section 2, most of these quasars have also been stud-ied spectroscopically in the mid-infrared with Spitzer(Schweitzer et al. 2006, 2008; Netzer et al. 2007; Veilleuxet al. 2009a). In addition, nearly all of the quasars inthis sample are part of X-QUEST, an archival XMM-Newton and Chandra X-ray spectroscopic survey of theQUEST sample (Teng & Veilleux 2010; Columns 13-17in Table 1). VLT and Keck near-infrared spectroscopicdata exist for a number of these objects (Dasyra et al.2007).Optical and near-infrared images of these objects have

been obtained from the ground (Surace et al. 2001;Veilleux et al. 2002; Guyon et al. 2006) and with HST(Veilleux et al. 2006; Kim et al. 2008; Hamilton et al.2008; Veilleux et al. 2009b), providing photometric andmorphological measurements on both the quasars andhost galaxies (e.g., morphological type, quasar-to-hostluminosity ratio, strength of tidal features). Far-infraredphotometry obtained with the Herschel PACS instru-ment exists for all of these objects (Lani et al. 2017;Shangguan et al. 2018), while far-infrared spectra cen-tered on the OH 119 µm feature exist for five of them(Veilleux et al. 2013a). Finally, Green Bank Telescope(GBT) H I 21-cm line emission and absorption spectraare available for 16 of these quasars (Teng et al. 2013).

4 Veilleux et al.

The connection between UV and X-ray propertiesis critical, so we searched the literature for addi-tional X-ray measurements (ignoring older ones fromROSAT/ASCA). These are listed and referenced in Ta-ble 1. For Chandra observations from Teng & Veilleux

(2010) where no absorbing column was detected, morerecent constraints from Ricci et al. (2017) (based onSwift/BAT detections) are available in some cases. Inthese cases, we substitute the newer measurement of ab-sorbing column.

Highly Ionized N V and O VI Outflows in the QUEST Quasars 5

Tab

le1.

Prope

rtiesof

theQUEST

Qua

sars

intheSa

mple

Nam

eOther

zlogνLν(U

V)logR

Rad

ioα

OX

log(L

bol

L

)α

AG

Nlog(L

IRL

)log(M

BH

M

)logη

Edd

log(L

SX)

log(L

HX)

ΓX

NH

Ref.

Nam

e[erg

s−1]

Class

[erg

s−1]

[erg

s−1]

[102

2cm

−2]

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

PG

0007

+10

6IIIZw

20.08

9344

.55

+2.29

Flat−1.43

12.24

1a11

.63

8.07

+0.4

5−

0.4

6−0.34

+0.4

6−

0.4

543

.94

44.18

1.73

+0.0

4−

0.0

40.11

0.0

2−

0.0

1a,

4,5

PG

0026

+12

9...

0.14

5445

.32

+0.03

Quiet−1.50

12.08

0.98

6+0.0

14

−0.0

27

11.71

8.49

+0.4

4−

0.4

5−0.93

+0.4

5−

0.4

444

.40

2.00

+0.1

3−

0.1

1<0.01

g,5

PG

0050

+12

4IZw

10.05

8944

.24

−0.48

Quiet−1.56

12.08

0.92

5+0.0

75

−0.0

94

12.04

7.33

+0.6

2−

0.6

20.20

+0.6

2−

0.6

244

.04+

0.0

5−

0.0

943

.88+

0.0

5−

0.0

92.25

+0.0

5−

0.0

30.09

+0.0

8−

0.0

2f,6

43.78+

0.0

2−

0.1

243

.64+

0.0

1−

0.0

42.09

+0.0

3−

0.0

30.04

+0.0

3−

0.0

26

43.633

+0.0

05

−0.0

052.37

+0.0

8−

0.0

40.04

53

PG

0157

+00

1Mrk

1014

0.16

3345

.34

+0.33

Quiet−1.60

12.70

0.72

7+0.2

36

−0.2

71

12.67

8.06

+0.6

1−

0.6

2−0.01

+0.6

3−

0.6

543

.92+

0.0

4−

0.0

443

.83+

0.1

1−

0.2

52.54

+0.0

9−

0.0

9<0.00

9e,

3,6

44.00+

0.0

4−

0.0

843

.86+

0.0

4−

0.0

42.1+

0.1

−0.1

<0.00

93,

6

PG

0804

+76

1...

0.10

045

.54

−0.22

Quiet−1.52

12.09

0.99

6+0.0

04

−0.0

04

11.98

8.73

+0.4

3−

0.4

3−1.16

+0.4

3−

0.4

344

.54

44.45

2.27

+0.0

9−

0.2

00.04

4+0.0

07

−0.0

10

a

PG

0838

+77

0VII

Zw

244

0.13

2444

.83

−0.96

Quiet−1.54

11.77

0.94

5+0.0

27

−0.0

18

11.66

8.05

+0.6

1−

0.6

2−0.82

+0.6

2−

0.6

144

.152

+0.0

09

−0.0

14

43.54+

0.0

4−

0.0

51.49

+0.0

8−

0.0

8<0.1b

d,4,

5

PG

0844

+34

9...

0.06

444

.59

−1.52

Quiet−1.54

11.45

0.97

1+0.0

29

−0.0

49

11.18

7.86

+0.4

6−

0.4

9−0.94

+0.4

9−

0.4

644

.152

+0.0

09

−0.0

14

43.80+

0.0

3−

0.0

32.66

+0.0

5−

0.0

66.13

+3.0

3−

1.3

9a,

6

PG

0923

+20

1...

0.19

245

.42

−0.85

Quiet−1.57

12.46

0.99

0+0.0

00

−0.0

00

12.05

7.90

+0.6

1−

0.6

20.04

+0.6

2−

0.6

1i

PG

0953

+41

4...

0.23

4145

.95

−0.36

Quiet−1.50

12.53

0.98

2+0.0

18

−0.0

18

12.20

8.33

+0.4

4−

0.4

4−0.33

+0.4

4−

0.4

445

.037

+0.0

06

−0.0

09

44.81+

0.0

2−

0.0

32.44

+0.0

3−

0.0

318

.52+

9.8

4−

5.6

h,6

PG

1001

+05

4...

0.16

1144

.93

−0.30

Quiet−2.13

11.87

0.83

6+0.1

23

−0.1

00

11.66

7.63

+0.6

1−

0.6

2−0.35

+0.6

2−

0.6

243

.00+

0.1

8−

0.3

043

.08+

0.1

5−

0.6

02.01

+0.6

7−

0.4

88.09

+5.4

7−

3.5

7e,

6

PG

1004

+13

04C

+13

.41

0.24

0645

.30

+2.36

Steep<−2.01

12.69

0.96

3+0.0

37

−0.0

36

12.22

9.16

+0.6

1−

0.6

2−1.01

+0.6

2−

0.6

143

.48+

0.0

5−

0.0

543

.76+

0.0

8−

0.1

31.67

+0.2

0−

0.1

12.99

+2.6

7−

1.3

7e,

6

43.51+

0.0

4−

0.1

543

.89+

0.0

7−

0.2

21.52

+0.1

7−

0.2

61.44

+0.6

4−

0.6

96

PG

1116

+21

5...

0.17

6545

.79

−0.14

Quiet−1.57

12.55

0.99

1+0.0

09

−0.0

08

12.28

8.42

+0.6

1−

0.6

2−0.39

+0.6

2−

0.6

144

.927

+0.0

21

−0.0

06

44.65+

0.0

3−

0.0

42.53

+0.0

4−

0.0

327

.21+

16.0

1−

11.2

6h,

6

44.922

+0.0

04

−0.0

04

44.65+

0.0

3−

0.0

42.49

+0.0

1−

0.0

131

.61+

5.1

4−

4.1

36

44.93+

0.0

1−

0.0

144

.67+

0.0

3−

0.0

32.51

+0.0

4−

0.0

420

.21+

5.9

4−

5.1

06

PG

1126−04

1Mrk

1298

0.06

044

.29

−0.77

Quiet−2.13

11.53

0.96

2+0.0

38

−0.0

75

11.52

7.64

+0.6

1−

0.6

2−0.65

+0.6

2−

0.6

143

.04+

0.0

5−

0.0

543

.11+

0.0

5−

0.1

21.95

+0.1

0−

0.1

04.66

+0.4

2−

0.3

9a,

6

PG

1211

+14

3...

0.08

0944

.96

+1.39

Steep−1.57

11.97

1.00

0+0.0

00

−0.0

00

11.74

7.85

+0.6

1−

0.6

2−0.40

+0.6

2−

0.6

144

.328

+0.0

05

−0.0

07

43.94+

0.0

1−

0.0

12.83

+0.0

2−

0.0

212

.98+

0.9

4−

0.9

0h,

6

44.201

+0.0

05

−0.0

05

43.89+

0.0

1−

0.0

12.63

+0.0

2−

0.0

212

.40+

1.6

4−

1.4

96

PG

1226

+02

33C

273

0.15

846

.50

+3.06

Flat−1.47

13.03

0.94

9+0.0

51

−0.1

28

12.80

8.41

+0.1

5−

0.2

40.08

+0.2

4−

0.1

645

.491

+0.0

02

−0.0

0245

.742

+0.0

08

−0.0

082.07

+0.0

1−

0.0

1<0.01

a,5,

6

45.461

+0.0

04

−0.0

0445

.722

+0.0

05

−0.0

061.81

+0.0

1−

0.0

1<0.01

5,6

45.591

+0.0

02

−0.0

0245

.820

+0.0

04

−0.0

052.28

+0.0

1−

0.0

1<0.01

5,6

45.663

+0.0

02

−0.0

0145

.825

+0.0

06

−0.0

062.08

+0.0

1−

0.0

1<0.01

5,6

45.461

+0.0

03

−0.0

04

45.67+

0.0

1−

0.0

22.13

+0.0

2−

0.0

2<0.01

5,6

45.544

+0.0

03

−0.0

0345

.941

+0.0

10

−0.0

071.96

+0.0

1−

0.0

2<0.01

5,6

PG

1229

+20

4Mrk

771

0.06

444

.42

−0.96

Quiet−1.49

11.57

0.98

5+0.0

15

−0.0

30

11.27

7.76

+0.4

6−

0.4

8−0.71

+0.4

8−

0.4

643

.785

+0.0

07

−0.0

08

43.61+

0.0

2−

0.0

22.38

+0.0

3−

0.0

313

.52+

5.7

7−

3.3

6g,

6

Mrk

231

...

0.04

217

42.70

−1.92

12.61

0.70

9+0.0

66

−0.0

67

12.54

8.58

+0.5

0−

0.5

0−0.63

+0.5

0−

0.5

042

.13+

0.0

1−

0.0

442

.58+

0.0

1−

0.1

11.40

+0.0

3−

0.1

9.5+

2.3

−1.9

b,7,

8

19.4

+5.7

−4.4

7,8

PG

1302−10

2PKS13

02-102

0.27

8445

.83

+2.27

Flat−1.58

12.75

0.98

2+0.0

18

−0.0

37

12.49

8.77

+0.6

1−

0.6

2−0.55

+0.6

2−

0.6

144

.81

1.66

+0.1

0−

0.1

1<0.06

h,1

PG

1307

+08

5...

0.15

545

.35

−1.00

Quiet−1.52

12.35

0.95

2+0.0

48

−0.0

66

11.76

8.54

+0.4

4−

0.4

6−0.72

+0.4

6−

0.4

444

.02+

0.0

2−

0.0

244

.16+

0.0

5−

0.0

91.89

+0.1

1−

0.1

05.64

+2.6

2−

1.4

8a,

6

PG

1309

+35

5...

0.18

2945

.05

+1.26

Flat−1.71

12.32

0.87

0+0.1

30

−0.1

27

12.05

8.24

+0.6

1−

0.6

2−0.49

+0.6

2−

0.6

243

.87+

0.0

2−

0.0

143

.88+

0.0

4−

0.0

52.19

+0.0

7−

0.0

66.02

+3.6

8−

1.8

4i,6

PG

1351

+64

0...

0.08

8245

.22

+0.64

Quiet−1.78

12.05

0.77

9+0.1

43

−0.2

21

11.87

8.72

+0.6

1−

0.6

2−1.30

+0.6

2−

0.6

343

.398

+0.0

07

−0.0

23

43.23+

0.0

3−

0.0

42.42

+0.0

4−

0.0

414

.61+

5.7

2−

3.8

1h,

6

Tab

le1continued

6 Veilleux et al.Tab

le1(con

tinu

ed)

Nam

eOther

zlogνLν(U

V)logR

Rad

ioα

OX

log(L

bol

L

)α

AG

Nlog(L

IRL

)log(M

BH

M

)logη

Edd

log(L

SX)

log(L

HX)

ΓX

NH

Ref.

Nam

e[erg

s−1]

Class

[erg

s−1]

[erg

s−1]

[102

2cm

−2]

(1)

(2)

(3)

(4)

(5)

(6)

(7)

(8)

(9)

(10)

(11)

(12)

(13)

(14)

(15)

(16)

(17)

PG

1411

+44

2...

0.08

9644

.34

−0.89

Quiet−2.03

11.79

1.00

0+0.0

00

−0.0

00

11.66

8.54

+0.4

5−

0.4

6−1.27

+0.4

6−

0.4

543

.60+

0.0

5−

0.0

643

.41+

0.0

6−

0.1

82.41

+0.1

8−

0.1

526

.29+

3.7

6−

4.0

8h,

6

PG

1435−06

7...

0.12

945

.12

−1.15

Quiet−1.63

11.92

0.97

6+0.0

24

−0.0

40

11.47

8.26

+0.6

1−

0.6

2−0.86

+0.6

2−

0.6

144

.11+

0.0

4−

0.0

443

.94+

0.0

7−

0.0

82.36

+0.1

1−

0.1

0<0.1b

a,6

PG

1440

+35

6Mrk

478

0.07

745

.18

−0.43

Quiet−1.38

11.81

0.83

6+0.0

71

−0.0

81

11.76

7.36

+0.6

1−

0.6

2−0.15

+0.6

2−

0.6

144

.33+

0.0

1−

0.0

143

.74+

0.0

5−

0.0

63.02

+0.0

4−

0.0

48.92

+8.6

6−

3.5

1a,

6

44.375

+0.0

05

−0.0

04

43.90+

0.0

1−

0.0

12.86

+0.0

1−

0.0

114

.24+

2.8

8−

2.3

46

44.299

+0.0

06

−0.0

06

43.74+

0.0

3−

0.0

32.98

+0.0

2−

0.0

28.22

5+2.0

3−

1.4

96

44.127

+0.0

07

−0.0

08

43.64+

0.0

2−

0.0

32.86

+0.0

2−

0.0

212

.67+

3.4

5−

2.7

06

PG

1448

+27

3...

0.06

543

.78

−0.60

Quiet−1.59

11.44

0.99

7+0.0

03

−0.0

07

11.19

6.86

+0.6

1−

0.6

20.06

+0.6

2−

0.6

143

.949

+0.0

06

−0.0

07

43.49+

0.0

2−

0.0

22.80

+0.0

2−

0.0

116

.72+

6.2

4−

4.3

9a,

6

PG

1501

+10

6Mrk

841

0.03

644

.03

−0.44

Quiet−1.64

11.34

1.00

0+0.0

00

−0.0

00

11.13

8.42

+0.6

1−

0.6

2−1.59

+0.6

2−

0.6

144

.049

+0.0

05

−0.0

0243

.833

+0.0

08

−0.0

082.46

+0.0

2−

0.0

223

.12+

6.4

0−

4.5

6a,

6

44.090

+0.0

02

−0.0

0543

.845

+0.0

08

−0.0

072.50

+0.0

2−

0.0

218

.66+

3.6

0−

2.7

76

44.068

+0.0

02

−0.0

0443

.851

+0.0

07

−0.0

062.45

+0.0

2−

0.0

215

.88+

2.4

2−

2.0

76

43.740

+0.0

04

−0.0

0443

.672

+0.0

06

−0.0

062.26

+0.0

1−

0.0

113

.01+

0.5

7−

0.5

06

43.623

+0.0

04

−0.0

06

43.65+

0.0

1−

0.0

12.11

+0.0

2−

0.0

211

.48+

1.6

9−

1.5

66

PG

1613

+65

8Mrk

876

0.12

945

.43

+0.00

Quiet−1.21

12.30

0.82

0+0.1

06

−0.0

92

12.25

8.34

+0.4

6−

0.5

1−0.64

+0.5

1−

0.4

644

.28+

0.0

6−

0.0

344

.38+

0.0

7−

0.0

91.95

+0.1

0−

0.1

028

.24+

107.7

8−

20.6

7a,

6

44.44+

0.0

3−

0.0

244

.43+

0.0

5−

0.0

62.12

+0.0

8−

0.0

810

.45+

10.7

8−

4.7

76

PG

1617

+17

5Mrk

877

0.11

444

.93

−0.14

Quiet−1.64

11.75

0.90

3+0.0

97

−0.0

81

11.55

8.67

+0.4

4−

0.4

5−1.48

+0.4

5−

0.4

4a

PG

1626

+55

4...

0.13

345

.17

−0.96

Quiet−1.37

11.84

0.97

6+0.0

24

−0.0

19

10.90

8.39

+0.6

1−

0.6

2−1.08

+0.6

2−

0.6

144

.17+

0.0

8−

0.0

844

.21+

0.0

7−

0.0

72.04

+0.1

5−

0.1

4<0.01

a,5,

6

PG

2130

+09

9II

Zw

136

0.06

344

.46

−0.49

Quiet−1.47

11.78

0.99

5+0.0

05

−0.0

10

11.63

7.43

+0.4

3−

0.4

3−0.17

+0.4

3−

0.4

343

.708

+0.0

11

−0.0

09

43.62+

0.0

2−

0.0

22.29

+0.0

5−

0.0

55.91

+0.7

3−

0.6

2j,6

PG

2214

+13

9Mrk

304

0.06

5844

.45

−1.30

Quiet−2.02

11.78

0.99

8+0.0

02

−0.0

04

11.46

8.44

+0.6

2−

0.6

2−1.18

+0.6

2−

0.6

243

.46+

0.0

4−

0.0

743

.64+

0.0

5−

0.1

71.80

+0.1

6−

0.1

64.48

+0.6

8−

0.6

8e,

6

PG

2233

+13

4...

0.32

6545

.94

−0.55

Quiet−1.66

12.56

1a12

.33

7.93

+0.6

1−

0.6

20.12

+0.6

2−

0.6

144

.52

2.41

+0.1

8−

0.1

8<0.01

e,2

PG

2349−01

44C−01

.61

0.17

4245

.51

12.59

0.90

4+0.0

45

−0.0

54

11.90

9.14

+0.5

0−

0.5

0−1.11

+0.5

0−

0.5

044

.57

1.78

+0.2

0−

0.3

5<0.01

e,5

Tab

le1continued

Hig

hly

Ionized

NV

and

OV

IO

utflo

ws

inth

eQ

UEST

Quasa

rs

7Table 1 (continued)

Name Other z log νLν(UV) log R Radio αOX log(LbolL

) αAGN log(LIRL

) log(MBHM

) log ηEdd log(LSX) log(LHX) ΓX NH Ref.

Name [erg s−1] Class [erg s−1] [erg s−1] [1022 cm−2](1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)

Note—Column (1): Object name. Column (2): Other name. Column (3): Redshifts, with reference listed in Column (17). Where available,redshifts are based on the [O III] narrow line. For 3 quasars, we use H I (de Vaucouleurs et al. 1991; Carilli et al. 1998; Springob et al. 2005)instead; for two, full-spectrum fits to SDSS spectra (Schneider et al. 2010); and for a single case, CO data (Evans et al. 2006). Column (4):Logarithm of the monochromatic luminosity at rest-frame 1125 Å derived using the Galactic extinctions from Schlafly & Finkbeiner (2011) andthe reddening curve with RV = 3.1 of Fitzpatrick (1999). Column (5): logarithm of R, the ratio of radio-to-optical luminosity from Boroson& Green (1992). Column (6): Radio class − Quiet, Steep, or Flat depending on log R and radio spectral index from Boroson & Green (1992).Column (7): X-ray to optical spectral index αOX = 0.372 log (f2 keV/f3000 A) from Brandt et al. (2000), where f2 keV and f3000 A are therest-frame flux densities at 2 keV and 3000 Å, respectively. (For Mrk 231, we report the value from Teng et al. (2014).) Column (8): Bolometricluminosity in solar units calculated from 7 × L(5100 Å) + LIR (Netzer et al. 2007), where L(5100 Å) is the continuum luminosity λLλ at 5100 Årest wavelength and LIR is the 1 − 1000 µm infrared luminosity listed in column (10). We adopt a cosmology of H0 = 69.3 km s−1; Ωm = 0.287;Ωλ = 0.713 (WMAP9). Column (9): Fraction of the bolometric luminosity produced by the AGN, i.e. αAGN = LAGN/LBOL, based on the Spitzerresults (Veilleux et al. 2009a). The error bars are computed from the lowest and highest values among the six methods from this paper. Column(10): Logarithm of the 1 − 1000 µm infrared luminosity in solar units from Zhuang et al. (2018), except Mrk 231 (U et al. 2013), PG 1626+554(Lyu et al. 2017), and PG 2349−014 (Veilleux et al. 2009b). Column (11): Logarithm of the black hole mass in solar units from reverberationmapping (RM) measurements from The AGN Black Hole Mass Database (Bentz & Katz 2015) or, if unavailable, single-epoch measurements fromVestergaard & Peterson (2006), normalized down from f = 5.5 (Onken et al. 2004) to f = 4.3 to match RM scaling. According to Vestergaard &Peterson (2006), single-epoch measurements should have an extra 0.43 error added in quadrature. For PG 2349−014, we used the more uncertainphotometric measurement of Veilleux et al. (2009b). For 3C 273, we recorded the GRAVITY measurement (Gravity Collaboration et al. 2018).Column (12): Logarithm of the ratio of the bolometric luminosity to the Eddington luminosity. Column (13): Luminosity in the soft X-rays (0.5 −2 keV). Column (14): Luminosity of the hard X-rays (2 − 10 keV). Column (15): Photon index of the best-fit absorbed power-law distribution tothe X-ray emission (dN/dE ∝ E−ΓX where E is the X-ray photon energy). Column (16): Column density in units of 1022 cm−2. For most datafrom Teng & Veilleux (2010), this is from the best-fit absorbed power-law. Where this best fit is unabsorbed and Swift/BAT data are available,we substitute NH from Ricci et al. (2017). For X-ray related quantities, different rows = different observations, dates. Column (17): Referencesfor the redshift and X-ray measurements.

References—Redshift: (a) Boroson & Green 1992; (b) Carilli et al. 1998; (c) de Vaucouleurs et al. 1991; (d) Evans et al. 2006; (e) Hewett & Wild2010; (f) Ho & Kim 2009; (g) Hu et al. 2020; (h) Marziani et al. 1996; (i) Schneider et al. 2010; (j) Springob et al. 2005; X-ray: (1) Inoue et al.2007; (2) Jin et al. 2012; (3) Laha et al. 2018; (4) Piconcelli et al. 2005; (5) Ricci et al. 2017; (6) Teng & Veilleux 2010; (7) Teng et al. 2014; (8)Veilleux et al. 2014; (9) Waddell & Gallo 2020

aNo measurement; we assume a value of 1.

b Best-fit value is 0; we assume an upper limit of 1021 cm−2.

8 Veilleux et al.

0.0 0.1 0.2 0.3

11.5

12.0

12.5

13.0

0.0 0.1 0.2 0.3Redshift

11.5

12.0

12.5

13.0

log(

L bol/L

O •)

Figure 1. Bolometric (AGN + starburst) luminosities of theQUEST quasars in the sample as a function of their redshifts.

4. HST DATA

We obtained high-quality spectra for 19 quasarsusing the Cosmic Origins Spectrograph (COS) withgrating G130M under HST PID 12569 in Cycle 19(PI Veilleux). We acquired multi-epoch COS/G130Mdata on PG 1411+442 under programs 13451, 14460,and 14885 (PI Hamann). We searched for archivalCOS/G130M spectra of other quasars in the QUESTsample, as well as for multi-epoch data on the subsamplewe observed. We found archival data on 14 additionalQUEST quasars and recent multi-epoch exposures forthree quasars in the subsample first observed in Cycle19. We list the characteristics of these observations inTable 2.Since most of the Cycle 19 COS data have not yet

been the subject of a paper (the exceptions are Mrk 231and PG 1411+442; Veilleux et al. 2013b, 2016; Hamannet al. 2019b), we briefly summarize here how they wereobtained. A total of 24 orbits were allocated for these19 targets with most targets requiring one orbit. Theexceptions are PG 1004+130 (2 orbits) and Mrk 231(5 orbits). All but Mrk 231 are point sources withaccurate positions; they were acquired directly usingACQ/PEAKXD and ACQ/PEAKD. For Mrk 231, aNUV image was obtained with ACQ/IMAGE. All obser-vations were observed in time-tag mode to allow us to ex-clude poor quality data and improve thermal correctionand background removal. We split the exposures intofour segments of similar durations at two FP_POS set-tings (#2 and #4) and two wavelength settings (CEN-

WAVE) separated by ∼20 Å. This observing strategyreduces the fixed pattern noise and fills up the chip gapwithout excessive overheads.The observations include at least 1150–1450 Å in the

observer’s frame. This range includes redshifted O VIλλ 1032, 1038, N V λλ1238, 1243, Lyα λ1216 and/orLyβ λ1025 in emission and/or absorption. In at leasttwo cases (PG 1126−041, PG 1411+442, and perhapsalso PG 1001+054 and PG 1004+130; see Sec. 7.4), theweaker P V λλ1117, 1128 absorption lines are also de-tected. The specific lines covered depend on the quasarredshift. The short-wavelength cutoff of the COS pre-vents us from searching for O VI systems in quasars withz . 0.11, while N V systems are redshifted out of theCOS data in quasars with z & 0.18. It is therefore pos-sible to study both O VI and N V only over a limitedrange of quasar redshifts. Nevertheless, we achieve ourscience goals by covering at least one H I Lyman seriesline and one high-ionization doublet (O VI and/or N V).In at least two cases (PG 1411+442 and PG 1004+130),weaker and/or lower-ionization lines, such as C II λ1335,C III λ977, N III λ990, O I λ1304, Si II λ1260, Si IIIλ1206, and Si IV λλ1394, 1403, are also present in thespectra. These lines may be used to help constrain thelocation, ionization, total column densities (NH) andmetal abundances in the absorbing gas (e.g. Hamannet al. 2019b). PG 1004+130, one of the highest-redshiftsources in our sample, also shows S VI λλ933, 945.All of our sample have data with CENWAVE of 1291,

1300, 1309, 1318, and/or 1327, and almost all of thedata (with the exception of the final round of data onPG 1001+054) were obtained in COS lifetime positions(LP) 1–3. For these CENWAVE settings, the spectralresolution of COS increases with wavelength and hasdegraded somewhat with changes in LP, but is still >104

at all wavelengths. This corresponds to resolution betterthan 30 km s−1 FWHM at all wavelengths, ranging up toa peak of ∼15 km s−1 at LP1 and 1450 Å. Three quasarshave additional data from CENWAVE 1055, 1096, or1222 observations. For these CENWAVE values, thespectral resolution peaks in the FUVB (blue) segmentat >104, but is lower in FUVA, with average values of∼3000, 5000, and 104, respectively.We downloaded all exposures from the Hubble Legacy

Archive and determined that they were processed byCALCOS v3.3.10. For each quasar, we coadded all ex-posures with CENWAVE 1222–1327 into a single spec-trum using v3.3 of coadd_x1d.pro (Danforth et al. 2010),setting BIN=3. The resulting median S/N per binnedpixel over 1290–1310 Å is 11.5, with a standard devia-tion of 10.9 and a range of 2–52. (The low end of therange arises in a strong N V BAL in PG 1126−041.) We

Highly Ionized N V and O VI Outflows in the QUEST Quasars 9

Table 2. Summary of the Observations

Name Range [Å] CENWAVE [Å] texp[s] Date PID PI

(1) (2) (3) (4) (5) (6) (7)

PG 0007+106 1151–1470 1309/1327 1868 2011-12-14 12569 S. VeilleuxPG 0026+129 1133–1451 1291/1309 1868 2011-10-25 12569 S. VeilleuxPG 0050+124 1151–1470 1309/1327 1868 2012-11-01 12569 S. Veilleux

1133–1465 1291/1309/1327 7621 2015-01-20 13811 E. CostantiniPG 0157+001 1151–1469 1309/1327 1828 2012-01-25 12569 S. VeilleuxPG 0804+761 1136–1458 1291/1300/1309/1318 5510 2010-06-12 11686 N. AravPG 0838+770 1136–1458 1291/1300/1309/1318 8865 2009-09-24 11520 J. GreenPG 0844+349 1151–1470 1309/1327 1900 2012-03-06 12569 S. VeilleuxPG 0923+201 1133–1451 1291/1309 1860 2012-03-14 12569 S. VeilleuxPG 0953+414 1136–1458 1291/1300/1309/1318 4785 2011-10-18 12038 J. GreenPG 1001+054 1066–1367 1222 2068 2014-04-04 13423 R. Cooke

1140–1455 1291/1300/1309/1318 3165 2014-06-19 13347 J. Bregman1131–1429 1291 2902 2019-03-26 15227 J. Burchett

PG 1004+130 1133–1451 1291/1309 4107 2011-12-21 12569 S. VeilleuxPG 1116+215 1136–1458 1291/1300/1309/1318 4677 2011-10-25 12038 J. GreenPG 1126−041 1152–1470 1309/1327 1856 2012-04-15 12569 S. Veilleux

901–1200 1055 1874 2014-06-01 13429 M. Giustini1171–1467 1327 1580 2014-06-01 13429 M. Giustini900–1200 1055 1874 2014-06-12 13429 M. Giustini1171–1467 1327 1580 2014-06-12 13429 M. Giustini900–1200 1055 1874 2014-06-28 13429 M. Giustini1171–1467 1327 1580 2014-06-28 13429 M. Giustini901–1200 1055 1837 2015-06-14 13836 M. Giustini1171–1468 1327 1540 2015-06-14 13429 M. Giustini

PG 1211+143 1171–1472 1327 2320 2015-04-14 13947 J. LeePG 1226+023 1135–1470 1291/1300/1309/1318/1327 4002 2012-04-22 12038 J. GreenPG 1229+204 1152–1469 1309/1327 1868 2012-04-26 12569 S. Veilleux

Mrk 231 1152–1472 1309/1327 12536 2011-10-15 12569 S. VeilleuxPG 1302−102 1136–1458 1291/1300/1309/1318 5979 2011-08-16 12038 J. GreenPG 1307+085 1152–1470 1309/1327 1836 2012-06-16 12569 S. VeilleuxPG 1309+355 1133–1451 1291/1309 1896 2011-12-06 12569 S. VeilleuxPG 1351+640 1152–1470 1309/1327 2108 2011-10-21 12569 S. VeilleuxPG 1411+442 1152–1470 1309 1936 2011-10-23 12569 S. Veilleux

941–1241 1096 4954 2015-02-12 13451 F. Hamann1152–1453 1309 1917 2015-02-12 13451 F. Hamann941–1241 1096 2407 2016-04-16 14460 F. Hamann1152–1453 1309 1954 2016-04-16 14460 F. Hamann941–1241 1096 1783 2017-06-10 14885 F. Hamann1152–1453 1309 1847 2017-06-10 14885 F. Hamann

PG 1435−067 1133–1451 1291/1309 1864 2012-02-29 12569 S. VeilleuxPG 1440+356 1152–1470 1309/1327 1924 2012-01-26 12569 S. VeilleuxPG 1448+273 1136–1448 1291/1309 2946 2011-06-18 12248 J. TumlinsonPG 1501+106 1132–1434 1291 3121 2014-07-06 13448 A. FoxPG 1613+658 1145–1467 1300/1309/1318/1327 9499 2010-04-08 11524 J. Green

1133–1429 1291 3080 2010-04-09 11686 N. AravPG 1617+175 1133–1451 1291/1309 1844 2012-06-16 12569 S. VeilleuxPG 1626+554 1136–1458 1291/1300/1309/1318 3318 2011-06-15 12029 J. GreenPG 2130+099 1135–1458 1291/1300/1309/1318 5513 2010-10-28 11524 J. GreenPG 2214+139 1152–1463 1309/1327 1401 2011-11-08 12569 S. Veilleux

1138–1434 1291 2082 2012-09-21 12604 A. FoxPG 2233+134 1171–1472 1327 2104 2014-06-18 13423 R. CookePG 2349−014 1152–1470 1309/1327 1844 2011-10-20 12569 S. Veilleux

Note—Column (1): Name of object; Column (2): Wavelength range, in Å; Column (3): CENWAVE setting(s);Column (4): Exposure time, in seconds; Column (5): Start date; Column (6): Proposal ID; Column (7):Program principal investigator.

separately coadded the two quasar datasets with CEN-WAVE 1055 and 1096.

5. DATA ANALYSIS

We conducted a uniform analysis of the high-ionization absorbers in our sample. In this section, wedescribe the methods we used to identify and character-ize these absorption features.We used v0.5 of the publicly available, IDL-based IFS-

FIT package (Rupke 2014; Rupke & Veilleux 2015) to

model the absorption lines. The rest of the softwareused to model the data (continuum fitting, plotting,regressions) is contained in or called from our publicCOSQUEST repository on GitHub (Rupke 2021a).

5.1. Model Fitting

The starting point of the analysis is to identify thevarious emission and absorption lines produced by thequasars and their environments. Since our program isfocused on QSO and ULIRG outflows, we only identify

10 Veilleux et al.

and measure absorption lines within ∼10,000 km s−1 ofthe QSO redshifts. We refer to these lines as “associ-ated” absorbers. Identifications of the foreground “inter-vening” absorbers can be found in Tripp et al. (2008),Savage et al. (2014), and Danforth et al. (2016). We firstcompare each quasar spectrum against a list of commonUV absorbers in quasar spectra (Prochaska et al. 2001,Figure 2). We list the quasar redshifts in Table 1. Mostof these redshifts are derived from narrow optical emis-sion lines ([O III], Hβ) and may underestimate the truerecession velocities since some fraction of the line emis-sion may arise from the outflowing material itself (e.g.Rupke et al. 2017). See Teng et al. (2013) for a compar-ison of these measurements with the H I 21-cm emissionand absorption line profiles.

PG1126−041, z = 0.060

1200 1250 1300 1350 1400 1450Observed Wavelength (Å)

0.0

5.0

10.0

15.0

Fλ/1

0−14 (

ergs

s−

1 cm

−2 Å

−1 )

1160 1170 1180 1190 12000.00.51.01.52.02.53.03.5

PII

1152

CI 1

157

FeI

I 109

6

PV

111

7

FeI

I 112

1F

eIII

1122

SIII

119

0, S

iII 1

190

FeI

I 112

5S

iII 1

193

PV

112

8

NI 1

199

NI 1

200

NI 1

200

NI 1

134,

NI 1

134

NI 1

134

1210 1220 1230 1240 12500.0

1.0

2.0

3.0

SiII

I 120

6

FeI

I 114

3

FeI

I 114

5

Lyα

PII

1152

CI 1

157

NV

123

8

CI 1

139

NV

124

2

SII

1250

SII

1253

1260 1270 1280 1290 1300 13100.02.04.06.08.0

10.012.014.0

SII

1259

SiII

126

0, F

eII 1

260

SIII

119

0, S

iII 1

190

SiII

119

3

NI 1

199

NI 1

200

NI 1

200

CI 1

277

SiII

I 120

6

CI 1

280

Lyα

PII

1301

, OI 1

302

SiII

130

4

1320 1330 1340 1350 13600.01.02.03.04.05.0

NV

123

8

CI 1

139

NiII

131

7, N

V 1

242

SII

1250

CI 1

328,

SII

1253

CII

1334

SII

1259

CII*

133

5S

iII 1

260,

FeI

I 126

0

CI 1

277

OI 1

355

CI 1

280

1370 1380 1390 1400 14100.01.0

2.0

3.0

4.0

NiII

137

0

PII

1301

OI 1

302

SiII

130

4

SiIV

139

3

NiII

131

7

SiIV

140

2

CI 1

328

CII

1334

CII*

133

5

1420 1430 1440 1450 14600.01.0

2.0

3.0

4.0

OI 1

355

NiII

137

0

NiII

145

4

NiII

146

7N

iII 1

467

Figure 2. An example of a FUV spectrum used in the study.Shown here is the COS spectrum of PG 1126−041, wherethe data are displayed in black and the expected positionsof the features in the Milky Way and quasar rest-frames areindicated in blue and red, respectively.

Next, we fit the continuum and broad line emission(Lyα, N V, and O VI) in three separate spectral win-dows around Lyα, N V, and O VI+Lyβ. In the twoquasars in which we fit P V, this continuum region isalso fit separately. Within each of these windows, weuse a piecewise function of 1–4 segments in the major-ity of cases. In relatively featureless spectral regions,these segments are low-order polynomials. In more com-

plex spectral regions we employ cubic B-splines. TheB-splines are themselves piecewise polynomials, and weseparate the spline knots by a typical interval of 3 Å. Weinvoke BSPLINE_ITERFIT from the SDSS IDLUTILSlibrary to fit the B-splines.In seven cases, the fits with piecewise functions are

poorly constrained. For PG 1001+054, PG 1004+130,PG 1411+442, and PG 1617+175, PG2130+099, andPG2214+139 this is due to broad, deep absorption fea-tures over which it is difficult to fit polynomials orsplines. For a seventh quasar—PG 1351+640—the poorconstraints are due to several narrow absorbers near thepeak of Lyα. In two of these cases, we instead use aLorentzian profile to fit Lyα. For the five others we usethe BOSS template from Harris et al. (2016), scale itmultiplicatively by a low-order power law, and add alinear pedestal.After fitting the continuum, we normalize the data in

each spectral window by dividing by this fit.We characterize the doublet absorption features (N V

λλ1238, 1243; O VI λλ1032, 1038; and P V λλ1117,1128) in the quasar spectra using simple model fits. Ourprimary objectives are to estimate the overall equivalentwidths and kinematics of the outflowing gas associatedwith these features (mass, momentum, and energy esti-mates are beyond the scope of the present paper, exceptfor a few special cases discussed in Section 7.4). Weare not aiming to derive precise column densities fromthe (often saturated) absorption line profiles, so the useof the precise COS line-spread function (LSF) is not re-quired here (we return to this point at the end of this sec-tion). If the lines within these doublets were unblended,fits to the intensity profiles of the individual lines wouldthus be sufficient. However, the doublet lines are oftenstrongly blended because of (1) strong blueshifts due tohigh outflow velocities and (2) broad line profiles dueto multiple clouds along the line of sight and/or largelinewidths. We thus adopt the doublet fitting procedureof Rupke et al. (2005), which is optimized for blendeddoublets. In this method, the total absorption profilesof a feature are fit as the product of multiple doubletcomponents. Each component is a Gaussian in opticaldepth τ vs wavelength with a constant covering factorCf . Within each doublet the two lines have a constantτ ratio. This allows us to simultaneously fit τ and Cf ,which are otherwise degenerate in the fit of a single line.The free parameters in the fit to each doublet compo-nent are thus Cf , peak τ , velocity width, and centralwavelength. The determination of the number of com-ponents needed in the fit is subjective and non-linear− it depends on the line complexity and data quality.The main goal here is to get a good fit to the absorption

Highly Ionized N V and O VI Outflows in the QUEST Quasars 11

PG1126−041, z=0.060

−5000 −4000 −3000 −2000 −1000 0 1000Velocity (km/s)

0.00.2

0.4

0.6

0.8

1.0

Nor

mal

ized

Fλ

PV 1117

0.00.2

0.4

0.6

0.8

1.0

PV 1128

0.00.2

0.4

0.6

0.8

1.0

OVI 1032

0.00.2

0.4

0.6

0.8

1.0

OVI 1038

0.00.2

0.4

0.6

0.8

1.0

NV 1238

0.00.2

0.4

0.6

0.8

1.0

NV 1242

0.00.2

0.4

0.6

0.8

1.0

Lyα

0.00.2

0.4

0.6

0.8

1.0

Lyβ

Figure 3. An example of interline comparison that is usedto identify absorbing systems associated with the quasars.The results shown here are for PG 1126−041, produced bydividing the spectrum shown in Fig. 2 by a smooth poly-nomical/spline/template fit to the continuum near the keyabsorption lines of our study and plotting the results in ve-locity space in the quasar rest frame. The data are in black,the components used to fit the absorption profiles are shownin blue, and the overall fit is shown in purple. The veloc-ity centroids of the main absorbing systems are indicated byblue (O VI) or red (Lyα, Lyβ, N V, P V) vertical dottedlines. The two strong lines in the panel labeled P V λ1128are Galactic ISM features without counterparts in P V λ1117or any of the other lines.

features to derive the equivalent widths and kinemat-

ics of the outflowing gas associated with these features.We do not attach a physical meaning to the individualcomponents in the fit.The general expression for the normalized intensity of

a doublet component is

I(λ) = 1− Cf + Cfe−τlow(λ)−τhigh(λ), (1)

where Cf is the line-of-sight covering factor (or the frac-tion of the background source producing the continuumthat is covered by the absorbing gas; though scatter-ing into the line of sight can also play a role) and τlow

and τhigh are the intrinsic optical depths of the lower-and higher-wavelength lines in the doublet (Rupke et al.2005). The background light source is assumed to bespatially uniform. The covering factor is the same forboth lines of the doublet. The peak (and total) opticaldepths of the resonant doublet lines in O VI, N V, andP V are related by a constant factor τlow/τhigh = 2.00

because of the 4-fold degeneracy in the upper state ofthe higher energy transition compared to the 2-fold de-generacy in the lower state. (The higher degeneracy isdue in turn to its higher total angular momentum quan-tum number j). For more than one doublet component,we use the product of the intensities of the individualcomponents, which is the partially-overlapping case ofRupke et al. (2005).Because the doublet profile shape–i.e., relative depths

of the two lines and trough shape–does not change sig-nificantly above optical depths τhigh of a few, we set alimit of τhigh ≤ 5. Out of 59 O VI components, 19 haveτhigh = 5, or 32%. For N V, 13 of 62 components haveτhigh = 5, or 21%.The results from these fits are also used to calculate

the total velocity-integrated equivalent widths of the ab-sorbers in the object’s rest frame,

Weq =

∫[1− f(v)]dv, (2)

the weighted average outflow velocity,

vwtavg =

∫v[1− f(v)]dv

Weq, (3)

and the weighted outflow velocity dispersion,

σrms =

(∫(v − vwtavg)2[1− f(v)]dv

Weq

) 12

, (4)

a measure of the second moment in velocity space of theabsorbers in each quasar. These quantities are similarto those defined by Trump et al. (2006), but withoutthe constraints on depth, width, or velocity. These con-straints have little effect on the results for our sample,

12 Veilleux et al.

but we find it useful to include possibly inflowing ab-sorbers. Note that Weq, vwtavg, and σrms are not cor-rected for partial covering. To test the impact of thisassumption on our results, we have recomputed themafter changing the absorption lines so that they haveCf = 1 instead of the measured Cf , and then redid theregression analysis discussed in Sec. 5.2. Only very smallchanges of order 1% in the p-values are observed if wecorrect for partial covering.Figure 3 shows the fits to the spectrum presented in

Figure 2. The fits to all of the features detected in theFUV spectra of the 33 quasars in our sample are pre-sented in Appendix A, and the results derived from thesefits are tabulated in Table 3.We computed errors in best-fit parameters and derived

model quantities by refitting the model spectrum 1000times. In each case we added Gaussian-distributed ran-dom errors to each pixel in the model with σ equal to themeasurement error. These formal errors are small dueto the high S/N in our data. Errors due to continuumplacement are likely to dominate the true error budget.We also estimated upper limits to the doublet equiva-

lent width in cases where we did not detect N V and/orO VI. To do so, we assumed an optically-thick (τ1243

or τ1038 = 5), v = 0, σ = 50 km s−1 absorption line.We set the covering factor equal to half the root-mean-square deviation in the continuum within ±0.5 Å of theexpected rest-frame location of each line in the doublet.(The factor-of-2 accounts for fitting 2 lines instead of 1.)We set the limit equal to the resulting model equivalentwidth.The optical depths and covering factors derived from

our fitting scheme are approximations. Though itis a physically-motivated way to decompose strongly-blended doublets, the method implicitly assumes thatthe velocity dependences of Cf and τ can be describedas the sum of discrete independent Gaussians. In re-ality, they are probably more complex functions of ve-locity (e.g. Arav et al. 2005, 2008). In several cases–the N V absorbers in PG 1001+054, PG 1411+442,PG 1617+175, and PG2214+139, and the O VI ab-sorbers in PG 1001+054 and PG 1004+130–the fitsinclude very broad components that cannot be distin-guished from complexes of narrower lines given the dataquality. In two O VI absorbers (PG 0923+201 andPG 1309+355), there are no data on the blue line be-cause it is contaminated by geocoronal Lyα, so any con-straints on τ and Cf come solely from line shape. Fi-nally, in four O VI fits (PG 1001+054, PG 1004+130,PG 1126−041, and PG 1617+175), the Lyβ and O VIabsorption lines blend together and cannot be easilyseparated in the fit. In three of these cases (all but

PG 1001+054), we simply fit the visible absorption asdue solely to O VI at wavelengths in which there is atleast some O VI absorption contributing to the spec-trum. For the fourth case, we are able to roughly sep-arate the lines by fitting only down to a specific wave-length. A detailed object-by-object discussion is givenin Appendix A.Despite these caveats, the fitting procedure is suffi-

cient to meet our primary objectives of estimating theoverall equivalent widths and kinematics of these fea-tures. The 3-σ detection limit on the doublet equivalentwidths is typically ∼ 20 mÅ in our data although itvaries from one spectrum to the other.We have conducted detailed tests of the impact of the

COS LSF on our measurements to verify that the useof the precise COS LSF is not required here. In oneseries of simulations, we created a series of fake, sat-urated Voigt line profiles with a median S/N of 5 perpixel and line widths ranging from σ = 10 km s−1 to 50km s−1. We convolved the profiles with the LSF down-loaded from the COS website. We find that the LSFcauses a difference of up to only ∼10% on the line widthand covering fraction measurements for lines with σ ≥20 km s−1 (the corresponding Doppler b parameter ofthe Voigt profile is

√2 σ ' 28 km s−1), which is smaller

than the values measured for nearly all of the absorbersdetected in our objects (Table 3). We have also run aCOS LSF analysis on a ULIRG with narrow N V absorp-tion features, taken from the sample of Paper II. Usingthe method described here, we get a Doppler b param-eter of 78 km s−1 and covering fraction of 0.84, whilethe COS LSF gives 83 km s−1 and 0.82, respectively,confirming that the results for the relatively broad ab-sorbers reported in the present paper are reliable.

5.2. Regressions

To search for connections between outflow andquasar/host properties, we computed linear regressionsbetween the properties in Tables 1 and 3. In mostcases, we apply the Bayesian model in LINMIX_ERR(Kelly 2007). We use the Metropolis-Hastings samplerand a single Gaussian to represent the distribution ofquasar/host parameters (except for Weq vs. AGN frac-tion, for which we used NGAUSS=3). LINMIX_ERRpermits censored y-values, which is the case for Weq.When we compute the regressions for the indepen-

dent variable NH, however, the x-axis values are alsocensored. In this case we turn to the method of Isobeet al. (1986) for computing the Kendall tau correlationcoefficient with censored data in both axes. We use theimplementation of pymccorrelation (Privon et al. 2020),which in turn perturbs the data in Monte Carlo fash-

Highly Ionized N V and O VI Outflows in the QUEST Quasars 13

Table 3. Results from the Multi-Component Fits to the Absorbers

Name Line Weq vwtavg σrms # comp.

Å km s−1 km s−1

(1) (2) (3) (4) (5) (6)

PG0007+106 N V <0.16PG0026+129 N V <0.09

O VI <0.12PG0050+124 N V 0.88+0.019

−0.017 -1106.3+19.0−18.7 595.0+18.0

−16.4 6PG0157+001 N V <0.13

O VI <0.09PG0804+761 N V 0.02+0.003

−0.003 591.0+1.8−1.7 12.3+1.9

−1.8 1O VI 0.18+0.009

−0.007 571.0+1.6−1.5 27.8+1.1

−1.1 1PG0838+770 N V <0.12

O VI <0.06PG0844+349 N V 0.73+0.011

−0.011 151.4+0.8−0.8 31.8+0.5

−0.5 2PG0923+201 O VI 4.48+0.072

−0.072 -3048.3+11.4−11.9 335.6+9.3

−8.7 1PG0953+414 O VI 0.20+0.006

−0.006 -825.4+14.7−13.9 418.3+7.2

−7.1 2PG1001+054 N V 6.07+0.058

−0.059 -5969.9+24.0−37.4 1326.5+18.0

−11.7 4O VI 11.61+0.080

−0.072 -5743.9+33.2−83.1 1092.4+32.7

−27.6 6PG1004+130 O VI 24.29+0.245

−0.277 -5335.7+80.0−75.4 2970.2+72.3

−47.4 12PG1116+215 O VI 0.27+0.008

−0.008 -2271.6+43.2−46.3 908.7+22.1

−25.6 2PG1126-041 N V 10.66+0.037

−0.034 -2085.5+10.0−9.8 1076.1+6.9

−7.2 13O VI 16.80+1.891

−1.616 -2559.2+298.8−251.5 1482.2+132.6

−169.4 10P V 0.21+0.017

−0.017 -2234.2+5.9−6.1 48.2+4.3

−3.8 2PG1211+143 N V <0.05PG1226+023 N V <0.04

O VI <0.04PG1229+204 N V <0.09Mrk 231 N V <0.18PG1302-102 O VI <0.05PG1307+085 N V <0.10

O VI 0.15+0.017−0.016 -3406.2+11.9

−14.1 66.8+11.8−14.1 2

PG1309+355 O VI 8.57+0.038−0.031 -893.9+4.9

−4.7 364.5+3.2−3.1 7

PG1351+640 N V 4.36+0.036−0.034 -1264.4+10.0

−11.4 428.4+4.8−3.8 9

PG1411+442 N V 10.30+0.019−0.018 -1594.8+2.6

−2.4 562.7+2.5−2.3 4

P V 0.83+0.028−0.028 -1754.6+5.9

−5.5 131.0+3.8−3.6 2

PG1435-067 N V <0.15O VI <0.08

PG1440+356 N V 0.89+0.023−0.021 -1478.4+28.1

−26.7 775.3+27.0−25.8 3

PG1448+273 N V 3.22+0.038−0.033 -229.8+2.2

−2.1 164.1+1.0−1.0 4

PG1613+658 N V 0.14+0.005−0.004 -3714.6+6.1

−5.5 121.8+3.2−3.6 2

O VI 0.68+0.008−0.008 -3691.1+2.0

−1.8 126.6+0.8−0.8 2

PG1617+175 N V 3.00+0.059−0.055 -3094.7+19.9

−23.8 526.8+34.3−22.5 5

O VI 6.08+0.173−0.173 -3323.7+113.3

−133.0 920.2+83.2−71.6 8

P V 0.06+0.040−0.033 -3355.0+55.1

−38.0 42.1+46.1−25.8 1

PG1626+554 N V <0.08O VI <0.06

PG2130+099 N V 0.76+0.013−0.013 -1312.3+9.9

−9.6 540.3+9.5−9.3 3

PG2214+139 N V 8.01+0.024−0.023 -1461.1+17.9

−17.9 681.7+21.9−20.2 5

PG2233+134 O VI 0.17+0.025−0.024 -211.2+3.0

−3.0 17.3+3.2−2.9 1

PG2349-014 N V <0.12

Note—Column (1): Name of object. Column (2): N V means N V λ1238, 1243,O VI means O VI λ1032, 1038, and P V means P V λ1117, 1128. N V orO VI is not listed when it lies outside of the spectral range of the data. Column(3): Velocity-integrated equivalent widths (eq. 2). Column (4): Average depth-weighted outflow velocity (eq. 3), which is a measure of the average velocity ofthe outflow systems in each object. Column (5): Average depth-weighted outflowvelocity dispersion (eq. 4), which is a measure of the range in velocity of theoutflow systems in each quasar. Column (6): Number of absorption components.

14 Veilleux et al.

ion to compute the errors in the correlation coefficient(Curran 2014).For both regression methods, we computed the signifi-

cance of a correlation as the fraction of cross-correlationvalues r < 0 (r > 0) for a positive (negative) best-fit r.For LINMIX_ERR, the r values are draws from the pos-terior distribution, while for pymccorrelation they areresults of the Monte Carlo perturbations.We do not consider the N V and O VI points inde-

pendent for the purposes of the regressions. Therefore,where both doublets are present in the data for a givenquasar, we compute the average measurement (eitherdetection or limit) from the two lines. If only one lineis detected, we use that measurement rather than av-eraging a detection and a limit. Where multiple X-raymeasurements exist for a quasar, we take the average.Errors in LBOL, LIR/LBOL, LFIR/LBOL, and αOX areunknown, so for the purposes of regression we fix theerrors to 0.1 dex. For νLν(UV), we ignore the negligiblestatistical measurement errors.

6. RESULTS

The results from our spectral analysis of the HST spec-tra are summarized in Table 3. In this section, we inves-tigate whether the presence or nature of quasar-drivenoutflows and starburst winds correlate with the prop-erties of the quasars and host galaxies. The quantitiesthat we consider in our correlation matrix are listed inTable 1 and defined in the notes to that table. Theresults from the statistical and regression analyses aresummarized in Tables 4 and 5.Note that we do not make a distinction between

quasar-driven outflows and starburst-driven winds inthis section, and we only consider absorption lines within10,000 km s−1 of the QSO redshift (inclusion of lines atgreater displacements leads to unacceptable contamina-tion by intervening absorbers). A more comprehensiveassessment of the detected outflows is conducted in Sec-tion 7, after we have considered the line profiles morefully, including signs of saturation and/or partial cover-ing of the continuum source (Sec. 6.2) and the overallkinematics of the outflowing gas (Sec. 6.4). Similarly,the comparison of our results with those from previousstudies is postponed until Section 7, once the resultsfrom our spectral analysis have been fully presented.

6.1. Rate of Incidence of Outflows

Figures 4 and 5 show the median velocities in thequasar rest frame of all of the detected N V and O VIabsorption-line systems, sorted from top to bottom bydecreasing redshift and bolometric luminosity, respec-tively. The first of these figures clearly illustrates the

fact mentioned in Section 4 that our ability to detectthe N V and O VI features is limited by the spectralcoverage of the COS data to z . 0.18 and z & 0.11,respectively (systems outside of the spectral range areindicated by an “x” in this figure and Fig. 5).

0 −2000 −4000 −6000 −8000v50 (km/s)

PG1501+106

Mrk 231

PG0050+124

PG1126−041

PG2130+099

PG0844+349

PG1229+204

PG1448+273

PG2214+139

PG1440+356

PG1211+143

PG1351+640

PG0007+106

PG1411+442

PG0804+761

PG1617+175

PG1613+658

PG1435−067

PG0838+770

PG1626+554

PG0026+129

PG1307+085

PG1226+023

PG1001+054

PG0157+001

PG2349−014

PG1116+215

PG1309+355

PG0923+201

PG0953+414

PG1004+130

PG1302−102

PG2233+134

Qua

sars

, sor

ted

by z

NV OVI00

NV 1238,1243O VI 1032,1038NV, σ < 25 km/sO VI, σ < 25 km/s

not in spectral rangeaffected by geocoronal line or chip gapundetected

Figure 4. Median velocities of the N V and O VI absorbingsystems detected in the QUEST quasars of our sample. Notethat the faster outflows with more negative velocities lie onthe right in this figure. The objects are sorted from top tobottom in order of decreasing redshift. Red symbols markN V λλ1238, 1243 and blue symbols mark O VI λλ1032,1038. Open symbols indicate systems with velocity disper-sion σ < 25 km s−1. The two columns on the right indicatewhether N V (red) or O VI (blue) is within the spectralrange of the data (“x” indicates that it is not), affected bygeocoronal line or chip gap (encircled “x”), or simply unde-tected (downward-pointing triangle). A lack of symbol marksa detection.

A cursory examination of Figures 4 and 5 shows thatblueshifted N V or O VI absorption systems suggestiveof outflows (with equivalent widths above our 3-σ detec-tion limit of ∼ 20 mÅ) are detected in about 60% of thequasars in our sample, and there is no obvious trend inthe rate of incidence with redshift or bolometric lumi-nosity.The results of a more quantitative analysis based on β

distributions (Cameron 2011) are listed in Table 4. The

Highly Ionized N V and O VI Outflows in the QUEST Quasars 15

0 −2000 −4000 −6000 −8000v50 (km/s)

PG1501+106

PG1448+273

PG0844+349

PG1126−041

PG1229+204

PG1617+175

PG0838+770

PG2130+099

PG2214+139

PG1411+442

PG1440+356

PG1626+554

PG1001+054

PG1435−067

PG1211+143

PG1351+640

PG0050+124

PG0026+129

PG0804+761

PG0007+106

PG1613+658

PG1309+355

PG1307+085

PG0923+201

PG0953+414

PG1116+215

PG2233+134

PG2349−014

Mrk 231

PG1004+130

PG0157+001

PG1302−102

PG1226+023Q

uasa

rs, s

orte

d by

Lbo

l

NV OVI00

NV 1238,1243O VI 1032,1038NV, σ < 25 km/sO VI, σ < 25 km/s

not in spectral rangeaffected by geocoronal line or chip gapundetected

Figure 5. Same as Fig. 4, but the objects are sorted fromtop to bottom in order of decreasing bolometric luminosity.

overall rate of incidence of N V or O VI absorbers is61% with a 1-σ range of (52% − 68%), once taking intoaccount the spectral coverage of the data. This rate isvirtually the same for N V and O VI. Among quasarswith log LBOL/L > 12.0, this rate is 61% with a 1-σrange of (49% − 71%), while it is 60% (47% − 71%)among the systems of lower luminosities. These ratesare thus not significantly different from each other, andare similar to the rate of incidence of O VI outflows inlocal Seyfert 1 galaxies (Kriss 2004a,b) as well as C IV(Crenshaw et al. 1999) or X-ray (Reynolds 1997; Georgeet al. 1998) absorption.We have also searched for trends between the rate

of incidence of outflows and several other quantities.The detection rate of outflows (79%) among quasarsthat have strongly absorbed X-ray continua (NH > 1022

cm−2) is significantly higher than those that do not(25%) (Table 4). These rates differ at the 2-σ level(95.4%), where the ranges of the incidence rate are 56− 92% for quasars with absorbed X-ray continua and 9− 54% for the others. Using the scipy.stats imple-mentation of the Fisher exact test, the null hypothesisthat galaxies with strongly- and weakly-absorbed X-raycontinua UV absorbers are equally likely to show N Vor O VI absorbers is rejected at the 99.2% level. The

Table 4. Rate of Incidence of Outflows

Line Detection Total Fraction (1-σ range)

(1) (2) (3) (4)

All Quasars

N V 13 27 0.48 (0.39 − 0.58)O VI 12 20 0.60 (0.49 − 0.70)Both 5 14 0.36 (0.25 − 0.50)Any 20 33 0.61 (0.52 − 0.68)

log LBOL/L ≥ 12.0

N V 4 12 0.33 (0.23 − 0.48)O VI 9 14 0.64 (0.50 − 0.75)Both 2 8 0.25 (0.16 − 0.44)Any 11 18 0.61 (0.49 − 0.71)

log LBOL/L < 12.0

N V 9 15 0.60 (0.47 − 0.71)O VI 3 6 0.50 (0.32 − 0.68)Both 3 6 0.50 (0.32 − 0.68)Any 9 15 0.60 (0.47 − 0.71)

NH > 1022 cm−2

N V 10 16 0.62 (0.50 − 0.73)O VI 8 8 1.00 (0.81 − 0.98)Both 3 5 0.60 (0.38 − 0.76)Any 15 19 0.79 (0.67 − 0.85)

NH ≤ 1022 cm−2

N V 2 10 0.20 (0.13 − 0.37)O VI 2 10 0.20 (0.13 − 0.37)Both 1 8 0.12 (0.08 − 0.32)Any 3 12 0.25 (0.17 − 0.41)

αox ≥ −1.6

N V 7 17 0.41 (0.31 − 0.53)O VI 6 12 0.50 (0.37 − 0.63)Both 2 9 0.22 (0.14 − 0.41)Any 11 20 0.55 (0.44 − 0.65)

αox < −1.6

N V 6 9 0.67 (0.49 − 0.78)O VI 6 7 0.86 (0.64 − 0.91)Both 3 4 0.75 (0.48 − 0.85)Any 9 12 0.75 (0.59 − 0.83)

Note—Column (1): Feature(s) used in the statisticalanalysis. “Both” means both N V and O VI doubletsand “Any” means either N V or O VI doublet orboth; Column (2): Number of objects with detectedoutflows; Column (3): Number of objects in totalwith the appropriate redshift; Column (4): Fractionof objects with detected outflows. The two numbersin parentheses indicate the 1-σ range (68% probabil-ity) of the fraction of objects with detected outflows,computed from the β distribution (Cameron 2011).

16 Veilleux et al.

Table 5. Linear Regression Results

y x N p r

(1) (2) (3) (4) (5)

Weq log(LBOL/L) 32 0.046 -0.34+0.19−0.17

Weq log[λL1125/erg s−1] 32 0.140 -0.22+0.21−0.18

Weq AGN fraction 32 0.094 0.78+0.19−0.61

Weq log(LAGN/L) 32 0.054 -0.33+0.19−0.16

Weq log(MBH/M) 32 0.289 -0.33+0.65−0.43

Weq Eddington Ratio 32 0.168 -0.42+0.44−0.38

Weq αOX 31 0.002 -0.62+0.17−0.13

Weq log(LIR/LBOL) 32 0.067 0.37+0.20−0.25

Weq log(LFIR/LBOL) 30 0.094 -0.29+0.22−0.19

Weq log[N(H)/cm−2] 30 <0.001 0.19+0.03−0.03

Weq Γ 30 0.143 0.25+0.20−0.23

Weq log[F (0.5 − 2 keV)/erg s−1 cm−2] 26 0.005 -0.54+0.18−0.14

Weq log[F (2 − 10 keV)/erg s−1 cm−2] 29 0.002 -0.55+0.18−0.14

Weq log[L(0.5 − 2 keV)/erg s−1] 26 0.072 -0.33+0.22−0.19

Weq log[L(2 − 10 keV)/erg s−1] 30 0.009 -0.51+0.20−0.15

Weq log[L(0.5 − 10 keV)/erg s−1] 26 0.051 -0.38+0.23−0.18

Weq log[L(0.5 − 2 keV)/L(0.5 − 10 keV)] 26 0.192 0.21+0.22−0.24

Weq log[L(0.5 − 10 keV)/LBOL] 26 0.211 -0.20+0.25−0.21

vwtavg log(LBOL/L) 20 0.138 -0.27+0.25−0.22

vwtavg log[λL1125/erg s−1] 20 0.283 -0.15+0.25−0.22

vwtavg αOX 20 0.103 0.31+0.21−0.24

vwtavg log(LIR/LBOL) 20 0.115 0.37+0.25−0.30

vwtavg log(LFIR/LBOL) 20 0.496 0.00+0.25−0.26

vwtavg AGN fraction 20 0.340 0.35+0.49−0.80

vwtavg log(LAGN/L) 20 0.150 -0.24+0.23−0.21

vwtavg log(MBH/M) 20 0.286 -0.31+0.56−0.42

vwtavg Eddington Ratio 20 0.414 0.13+0.53−0.59

vwtavg log[N(H)/cm−2] 18 0.034 -0.15+0.08−0.08

vwtavg Γ 18 0.011 0.67+0.14−0.22

vwtavg log[F (0.5 − 2 keV)/erg s−1 cm−2] 17 0.008 0.61+0.15−0.21

vwtavg log[F (2 − 10 keV)/erg s−1 cm−2] 17 0.016 0.57+0.16−0.23

vwtavg log[L(0.5 − 2 keV)/erg s−1] 17 0.087 0.36+0.22−0.26

vwtavg log[L(2 − 10 keV)/erg s−1] 18 0.264 0.17+0.25−0.28

vwtavg log[L(0.5 − 10 keV)/erg s−1] 17 0.211 0.23+0.25−0.28

vwtavg log[L(0.5 − 2 keV)/L(0.5 − 10 keV)] 17 0.006 0.69+0.13−0.20

vwtavg log[L(0.5 − 10 keV)/LBOL] 17 0.009 0.64+0.14−0.20

σrms log(LBOL/L) 20 0.238 0.18+0.23−0.25

σrms log[λL1125/erg s−1] 20 0.425 -0.04+0.24−0.24

σrms αOX 20 0.004 -0.55+0.20−0.15

σrms log(LIR/LBOL) 20 0.305 -0.16+0.32−0.30

σrms log(LFIR/LBOL) 20 0.433 -0.04+0.25−0.25

σrms AGN fraction 20 0.389 -0.13+0.55−0.58

σrms log(LAGN/L) 20 0.247 0.17+0.23−0.25

σrms log(MBH/M) 20 0.301 0.29+0.43−0.59

σrms Eddington Ratio 20 0.374 -0.18+0.58−0.49

σrms log[N(H)/cm−2] 18 0.086 0.10+0.08−0.07

σrms Γ 18 0.048 -0.46+0.26−0.20

σrms log[F (0.5 − 2 keV)/erg s−1 cm−2] 17 0.016 -0.56+0.22−0.16

σrms log[F (2 − 10 keV)/erg s−1 cm−2] 17 0.015 -0.55+0.22−0.16

σrms log[L(0.5 − 2 keV)/erg s−1] 17 0.066 -0.40+0.26−0.21

σrms log[L(2 − 10 keV)/erg s−1] 18 0.137 -0.28+0.26−0.23

σrms log[L(0.5 − 10 keV)/erg s−1] 17 0.120 -0.32+0.27−0.23

σrms log[L(0.5 − 2 keV)/L(0.5 − 10 keV)] 17 0.022 -0.54+0.24−0.18

σrms log[L(0.5 − 10 keV)/LBOL] 17 0.008 -0.61+0.21−0.15

Note—Column (1): Dependent variable (absorption line property). Column (2):Independent variable (quasar/host property). Column (3): Number of points.Column (4): p-value of null hypothesis (no correlation). Column (5): Correlationcoefficient and 1σ errors. Underlined entries under col. (2) indicate significantcorrelations with p-values below 0.05.

rate of incidence of outflows among quasars with a steepX-ray to optical spectral index (αOX < −1.6; 75%) isalso higher than those with a shallow index (55%), al-though the Fisher exact test shows that this differenceis not significant (p = 0.45). A similar dependence onthe X-ray properties of the quasars has been reported inseveral studies of higher luminosity quasars and lowerluminosity Seyfert 1 galaxies using C IV λλ1548, 1550as a tracer of warm ionized outflows. We return to thisresult in Sections 6.3 and 6.4, and 7.

6.2. Optical Depths and Covering Factors

The distributions of the N V λ1243 and O VI λ1038optical depths and covering factors derived from theindividual components in the multi-component fits arepresented as histograms in Figure 6.

<0.1 0.1−0.4 0.4−1.6 1.6−50.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7Log(τ1243 or τ1038) Distributions

<0.1 0.1−0.4 0.4−1.6 1.6−50.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 NVOVINV+OVI

NVOVINV+OVI

0.0−0.2 0.2−0.4 0.4−0.6 0.6−0.8 0.8−1.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7Covering Factor Distribution

0.0−0.2 0.2−0.4 0.4−0.6 0.6−0.8 0.8−1.00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7 NVOVINV+OVI

NVOVINV+OVI

Figure 6. Distributions of the optical depths (left) andcovering factors (right) of the individual components used tofit the profiles of the N V (orange), O VI (purple), or jointN V + O VI (pink) absorption features.

Again, we repeat that the optical depths and coveringfactors presented here are only approximations. Never-theless, it is clear from the left panel in Figure 6 thata significant fraction of the absorbing systems are af-fected by saturation effects (τ1243 or τ1038 > 1), thereforemaking the equivalent widths of the N V and O VI fea-tures unreliable indicators of the total column densitiesof highly ionized gas in many of these cases.The right panel of Figure 6 shows that the mode of

the distribution of covering factors is consistent withunity, but ∼50% of the N V and O VI absorbers onlypartially cover the FUV quasar continuum emission (+possibly the broad emission line region − BELR; Fig. 6),consistent with small clouds located relatively near thequasars. As described at the end of Sec. 5.1, emissioninfill of the absorption profiles associated with the broadwings of the COS LSF is negligible and thus does notaffect this conclusion. We return to this result in Section7.1.

6.3. Outflow Equivalent Widths

Highly Ionized N V and O VI Outflows in the QUEST Quasars 17

The velocity-integrated equivalent widths (Weq; eq. 2)of the outflow systems in each quasar are listed in Table3. They span a broad range from ∼ 25 Å down to 20mÅ, near our 3-σ detection limit.The equivalent widths of the outflows were compared

against the properties of the quasars and host galaxieslisted in Table 1. Some of the results are shown in Figure7. By and large, we do not find any significant trendsbetween Weq and any of the quasar and host properties,except with some of the quantities that are derived fromthe X-ray data (Table 5). Taken at face value, this re-sult is surprising since, for instance, it means that theequivalent width of the outflow is largely agnostic of theproperties of the central engine over a range of ∼ 1.5 dexin power (FUV, bolometric, or quasar-only luminosity),∼2.0 dex in Eddington ratio, and ∼2.5 dex in black holemass. The lack of correlations with the properties of thehosts is less surprising since the quasar sample spans arelatively narrow range of values in these quantities sothe lack of correlation with these quantities may be at-tributed to the lower dynamical range.Examples of trends between Weq and the X-ray prop-

erties of the quasars are shown in panels (e), (f), and (g)of Figure 7. In panel (e), the equivalent width of the out-flow decreases with increasing HX luminosity. Panel (f)in this figure illustrates the dependence of the rate of in-cidence of these outflows on the X-ray column densitiesalready pointed out in Section 6.1. The stronger highlyionized outflows with Weq & 1 Å are only present inquasars with X-ray column densities above ∼ 1022 cm−2.While it is a required condition for a strong outflow, it isnot a sufficient condition since most quasars with theseX-ray absorbing column densities show either weak out-flows in the FUV (Weq < 0.3 Å) or none at all. Panel(g) also shows a distinct trend for strong outflows withWeq & 1 Å among objects with αOX . −1.7. A similartrend is observed when normalizing the X-ray luminosi-ties to the bolometric luminosities (not shown), but dis-appears when considering only the X-ray slope (e.g. theSX/HX ratio or index of the best-fit absorbed power-lawdistribution to the X-rays; not shown). Similar resultshave been found when considering C IV outflows (e.g.Brandt et al. 2000; Laor & Brandt 2002; Baskin & Laor2005; Gibson et al. 2009a,b). We return to this issue inSection 7 below.

6.4. Outflow Kinematics

Figure 8 shows the distributions of the velocity cen-troids and dispersions (σ) of the various individual com-ponents that were used to fit the N V and O VI absorbersin the quasar sample. About half of all of the indi-vidual components have blueshifted (outflow) velocities

ï2.0 ï1.6 ï1.2 ï0.8 ï0.4 0.0 0.4ï2

ï1

0

1

2

ï2.0 ï1.6 ï1.2 ï0.8 ï0.4 0.0 0.4Eddington Ratio

ï2

ï1

0

1

2

log(

Weq

/ Å)

p=0.168 r=ï0.42ï0.38+0.44 N=32

11.5 12.0 12.5 13.0-2

-1

0

1

2

11.5 12.0 12.5 13.0log(Lbol/LO •)

-2

-1

0

1

2

log(

Weq

/ Å)

p=0.046 r=ï0.34ï0.17+0.19 N=32

20 21 22 23 24ï2

ï1

0

1

2

20 21 22 23 24log[ N(H, Xïray) / cmï2]

ï2

ï1

0

1

2

log(

Weq

/ Å)

p<0.001 r=0.19ï0.03+0.03 N=30

ï2.2 ï2.0 ï1.8 ï1.6 ï1.4 ï1.2ï2

ï1

0

1

2

ï2.2 ï2.0 ï1.8 ï1.6 ï1.4 ï1.2_ox

ï2

ï1

0

1

2

log(

Weq

/ Å)

p=0.002 r=ï0.62ï0.13+0.17 N=31

ï0.6 ï0.4 ï0.2 0.0ï2

ï1

0

1

2

ï0.6 ï0.4 ï0.2 0.0log(LIR/Lbol)

ï2

ï1

0

1

2

log(

Weq

/ Å)

p=0.067 r=0.37ï0.25+0.20 N=32

43 44 45 46ï2

ï1

0

1

2

43 44 45 46log[L(2ï10 keV)/erg sï1]

ï2

ï1

0

1

2

log(

Weq

/ Å)

p=0.009 r=ï0.51ï0.15+0.20 N=30

6.8 7.2 7.6 8.0 8.4 8.8 9.2-2

-1

0

1

2

6.8 7.2 7.6 8.0 8.4 8.8 9.2log(MBH/MO •)

-2

-1

0

1

2

log(

Weq

/ Å)

p=0.289 r=ï0.33ï0.43+0.65 N=32

0.7 0.8 0.9 1.0ï2

ï1

0

1

2

0.7 0.8 0.9 1.0AGN fraction

ï2

ï1

0

1

2

log(

Weq

/ Å)

p=0.094 r=0.78ï0.61+0.19 N=32(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 7. The velocity-integrated equivalent widths, Weq,of the outflow systems in the QUEST quasars are plotted asa function of the (a) bolometric luminosities, (b) AGN bolo-metric fractions, (c) Eddington ratios, (d) black hole masses,(e) hard X-ray (2 − 10 keV) luminosities, (f) X-ray absorb-ing column densities, (g) X-ray to optical spectral indices,and (h) ratios of the infrared luminosities to the bolometricluminosities. Red squares mark N V λλ1238, 1243 and bluecircles mark O VI λλ1032, 1038. Triangles indicate upperlimits (in one or both quantities). The regression results (p-values, correlation coefficients r with 1σ errors, and numberof points N ; Section 5) are shown at the top of each panel.The actual points used in the regression, in which N V andO VI quantities and/or X-ray measurements are averagedfor a given quasar, are shown in each inset panel. The solidpoints are detections, while the open points are censored val-ues in one or both quantities plotted.

that lie between [−2000, 0] km s−1 and have 1-σ widths

18 Veilleux et al.

less than 40 km s−1. In a blindly selected sample of O VIabsorbers, Tripp et al. (2008) similarly found that themajority of associated absorbers are within 2000 km s−1

of the QSO redshift (see their Figure 15). Likewise, theyfound that the O VI line widths are < 40 km s−1. Upto ∼10% of the individual components in the presentsurvey have redshifted velocities of up to a few × 100km s−1; some of them may be attributed to uncertainor systematically blueshifted systemic velocities derivedfrom the quasar emission lines (Sec. 5) rather than ac-tual inflows.

−8000 −6000 −4000 −2000 > 00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7Velocity (km/s) Distributions

−8000 −6000 −4000 −2000 > 00.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

to −6000 to −4000 to −2000 to 0

NVOVINV+OVI

NVOVINV+OVI

10−40 40−80 80−160 >1600.0

0.2

0.4

0.6

0.8Log[σ / km/s ] Distributions

10−40 40−80 80−160 >1600.0

0.2

0.4

0.6

0.8 NVOVINV+OVI

NVOVINV+OVI

Figure 8. Distributions of the median velocities (left) andvelocity dispersions (right) of the individual componentsused to fit the profiles of the N V (orange), O VI (purple),and joint N V + O VI (pink) absorption features.

More physically meaningful kinematic quantities arethe weighted average velocities and velocity dispersionsof the outflow systems in each object (eqs. 3 and 4). Ofthe 20 detected absorbers in Table 3, 8 (3) have weightedaverage outflow velocities (velocity dispersion) in excessof 2000 (1000) km s−1.We find in Table 5 that there is no distinct trend be-

tween the weighted outflow velocities and velocity dis-persions and the quasar and host properties, except forthe lack of outflows in X-ray unabsorbed quasars (Fig.10), pointed out in Sec. 6.1, and the larger weighted out-flow velocity dispersions among X-ray faint sources withαOX . −2. The lack of a correlation between outflowvelocities and the quasar luminosities seems at odds withthose from most previous C IV absorption-line studies(e.g. Perry & O’Dell 1978; Brandt et al. 2000; Laor &Brandt 2002; Ganguly et al. 2007; Ganguly & Broth-erton 2008; Gibson et al. 2009a,b; Zhang et al. 2014;Rankine et al. 2020) and other multi-wavelength analy-ses (e.g., references in Sec. 1 and Veilleux et al. 2020).We examine this issue in more detail in Section 7 below.

7. DISCUSSION

7.1. Origins of the Absorption Features

20 21 22 23 24

0

ï2000

ï4000

ï6000

20 21 22 23 24log[ N(H, Xïray) / cmï2]

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.034 r=ï0.15ï0.08+0.08 N=18

7.0 7.5 8.0 8.5 9.0

0

-2000

-4000

-6000

7.0 7.5 8.0 8.5 9.0log(MBH/MO •)

0

-2000

-4000

-6000

v wta

vg (k

m/s

)

p=0.286 r=-0.31-0.42+0.56 N=20

ï0.6 ï0.4 ï0.2 0.0

0

ï2000

ï4000

ï6000

ï0.6 ï0.4 ï0.2 0.0log(LIR/Lbol)

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.115 r=0.37ï0.30+0.25 N=20

ï2.0 ï1.8 ï1.6 ï1.4 ï1.2

0

ï2000

ï4000

ï6000

ï2.0 ï1.8 ï1.6 ï1.4 ï1.2_ox

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.103 r=0.31ï0.24+0.21 N=20

43 44 45 46

0

ï2000

ï4000

ï6000

43 44 45 46log[L(2ï10 keV)/erg sï1]

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.264 r=0.17ï0.28+0.25 N=18

ï1.5 ï1.0 ï0.5 0.0

0

ï2000

ï4000

ï6000

ï1.5 ï1.0 ï0.5 0.0Eddington Ratio

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.414 r=0.13ï0.59+0.53 N=20

11.5 12.0 12.5 13.0

0

-2000

-4000

-6000

11.5 12.0 12.5 13.0log(Lbol/LO •)

0

-2000

-4000

-6000

v wta

vg (k

m/s

)

p=0.138 r=-0.27-0.22+0.25 N=20

0.7 0.8 0.9 1.0

0

ï2000

ï4000

ï6000

0.7 0.8 0.9 1.0AGN fraction

0

ï2000

ï4000

ï6000

v wta

vg (k

m/s

)

p=0.340 r=0.35ï0.80+0.49 N=20(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 9. Same as Fig. 7 but the weighted average veloci-ties.

The blueshifted N V and O VI absorption features re-ported in Section 6 may have several origins: quasar-driven outflows, starburst-driven winds, tidal debrisfrom the galaxy mergers, and intervening CGM. Herewe do not consider contamination of the quasar spec-tra by young stars since none of them show the obvi-ous spectral signatures of young stars (e.g., narrow andshallow N V or O VI absorption troughs accompaniedby redshifted emission). This is only an issue amongstarburst-ULIRGs (e.g. Martin et al. 2015).Telltale signs that the detected lines are formed in a

quasar-driven outflow include (1) line profiles that are

Highly Ionized N V and O VI Outflows in the QUEST Quasars 19

20 21 22 23 240

1000

2000

3000

20 21 22 23 24log[ N(H, Xïray) / cmï2]

0

1000

2000

3000

m rm

s (km

/s)

p=0.086 r=0.10ï0.07+0.08 N=18

ï0.6 ï0.4 ï0.2 0.00

1000

2000

3000

ï0.6 ï0.4 ï0.2 0.0log(LIR/Lbol)

0

1000

2000

3000

m rm

s (km

/s)

p=0.305 r=ï0.16ï0.30+0.32 N=20

ï2.0 ï1.8 ï1.6 ï1.4 ï1.20

1000

2000

3000

ï2.0 ï1.8 ï1.6 ï1.4 ï1.2_ox

0

1000

2000

3000

m rm

s (km

/s)

p=0.004 r=ï0.55ï0.15+0.20 N=20

43 44 45 460

1000

2000

3000

43 44 45 46log[L(2ï10 keV)/erg sï1]

0

1000

2000

3000

m rm

s (km

/s)

p=0.137 r=ï0.28ï0.23+0.26 N=18

ï1.5 ï1.0 ï0.5 0.00

1000

2000

3000

ï1.5 ï1.0 ï0.5 0.0Eddington Ratio

0

1000

2000

3000

m rm

s (km

/s)

p=0.374 r=ï0.18ï0.49+0.58 N=20

7.0 7.5 8.0 8.5 9.00

1000

2000

3000

7.0 7.5 8.0 8.5 9.0log(MBH/MO •)

0

1000

2000

3000m

rms (

km/s

)p=0.301 r=0.29ï0.59

+0.43 N=20

0.7 0.8 0.9 1.00

1000

2000

3000

0.7 0.8 0.9 1.0AGN fraction

0

1000

2000

3000

m rm

s (km

/s)

p=0.389 r=ï0.13ï0.58+0.55 N=20

11.5 12.0 12.5 13.00

1000

2000

3000

11.5 12.0 12.5 13.0log(Lbol/LO •)

0

1000

2000

3000

m rm

s (km

/s)

p=0.238 r=0.18ï0.25+0.23 N=20(a) (b)

(c) (d)

(e) (f)

(g) (h)

Figure 10. Same as Fig. 7 but for the weighted velocitydispersions.

blueshifted, broad, and smooth compared to the ther-mal line widths (. 10 − 20 km s−1 for highly ionizedN4+, P4+, and O5+ ions at T ' 104.5−5.5 K), (2) lineratios within the multiplets N V λ1238/λ1243, O VIλ1032/λ1038, and P V λ1117/λ1128 that imply partialcovering of the quasar emission source, and (3) and largecolumn densities in these high-ionization ions (Hamannet al. 1997b,a; Tripp et al. 2008; Hamann et al. 2019b).N V is typically very weak or absent in intervening sys-tems (Werk et al. 2016). High N V/H I and O VI/H Iare also much higher in associated absorbers than in in-tervening systems (e.g., Tripp et al. 2008).

Among the 20 quasars with N V or O VI absorp-tion systems suggestive of outflows, 17 objects haveabsorption line profiles that meet the first of theabove criteria (the only exceptions are PG 0804+761,PG 0844+349, and PG 2233+134). Many of the quasarswith blueshifted N V or O VI absorption lines show N Vλ1238/λ1243 and/or O VI λ1032/λ1038 line ratios thatalso meet criteria #2 and #3 (Fig. 6). Mrk 231 doesnot formally meet these criteria (since it has no N Vabsorption line and O VI falls outside of the spectralrange of the data), but it shows all of the characteristicsof a FeLoBAL at visible and NUV wavelengths (and itsLyα line emission is highly blueshifted; Veilleux et al.2013b, 2016, and references therein), so we include ithere among those with quasar-driven outflows. So, over-all, at least 18 quasars in our sample have absorptionfeatures suggestive of quasar-driven outflows.In 15 of the 20 absorber detections, the velocity

widths, FWHMrms ≡ 2.355 σrms, are below the mini-mum of 2000 km s−1 generally used for BALs (Weymannet al. 1981, 1991; Hamann & Sabra 2004; Gibson et al.2009a,b), so they fall in the category of mini-BALs (500< FWHMrms < 2000 km s−1) or NALs (FWHMrms <

500 km s−1). Moreover, in many cases, the profiles arehighly structured rather than smooth, and thus do notmeet the “BAL-nicity” criterion to be true BALs.The weak and narrow redshifted absorption features

in PG 0804+761 and PG 0844+349 are good candidatesfor infalling tidal debris.

7.2. Location and Structure of the Mini-BALs7.2.1. Depths of the Absorption Profiles

The depths of the mini-BALs may be used to putconstraints on the location of the outflowing absorbers.The source of the FUV continuum in these quasars ispresumed to be the accretion disk on scale of ∼ few× 1015 cm (. 0.01 pc), where we used equation (6)in Hamann et al. (2019a) assuming an Eddington ratioηEdd = 0.1. But it is clear from the spectra that inmany cases (e.g., PG 1001+054, 1004+130, 1126−041,1309+355, 1351+640, 1411+442, 2214+139) the mini-BALs absorb not only the FUV continuum emission butalso a significant fraction of the Lyα, N V, and O VIline emission produced in the BELR. The gas producingthe mini-BALs must therefore be located outside of theBELR on scales larger than

rBELR = 0.1

(LAGN

1046 erg s−1

)1/2

pc (5)

(e.g. Kaspi et al. 2005, 2007; Bentz et al. 2013). Theradius of the outer boundary of the BELR, rout, is likelyset by dust sublimation (Netzer & Laor 1993; Baskin &

20 Veilleux et al.

Laor 2018). For gas densities of 105 − 1010 cm−3, Baskin& Laor (2018) derive

rout' rsubl ' (3− 6)rBELR (6)

= (0.3− 0.6)

(LAGN

1046 erg s−1

)1/2

pc, (7)

where graphite grains of size ∼ 0.05 µm is assumed (Fig.5 in Baskin & Laor 2018). Note that these values aresmaller than those in Barvainis (1987) and Veilleux et al.(2020), which are based on silicate grains and lower gasdensities (thus lower evaporation temperatures).More can be said about the structure of the absorbing

gas from the fact that the N V and O VI absorptionfeatures are optically thick (τ & 1 − 5; Fig. 6) but arenot completely dark. The covering factors derived fromthe multi-component fits to the N V and O VI mini-BALs range from 0.1 to 1, a direct indication that theabsorbing material is compact and spatially inhomone-neous. The often structured velocity profiles of the N Vand O VI mini-BALs (N V in PG 1411+442 is arguablythe only exception) also suggest a high level of kinematicsub-structures in the outflows, different from the smoothBALs observed in high-luminosity quasars. These prop-erties of the mini-BALs may indicate one of two things:(1) Our line of sight is not aligned along the directionof the outflowing stream of gas as in the case for theBALs (e.g. Murray et al. 1995; Elvis 2000; Ganguly et al.2001), but instead intercepts only a small fraction of thisstream and results in a covering factor of the backgroundemission that is highly dependent on the inhomogene-ity of the outflowing gas. (2) Another equally plausi-ble explanation is that (structured) mini-BALs form inmore sparse outflows or (in the unified outflow modeldiscussed for high-z quasars) in more sparse outflow re-gions, e.g., at higher latitudes above the disk.

7.2.2. Variability of the Absorption Profiles

Additional constraints on the location and structureof the BALs and mini-BALS in our sample may be ob-tained from profile variability. There is a vast literatureon this topic (e.g., Gibson et al. 2008; Hamann et al.2008; Gibson et al. 2010; Capellupo et al. 2012; FilizAk et al. 2012, 2013; Grier et al. 2015; He et al. 2019;Yi et al. 2019, and references therein). In our sample,four of the mini-BALs (PG 1001+054, PG 1126−041,PG 1411+442, and PG 2214+139) have been observedat two different epochs or more, and can therefore besearched for mini-BAL profile variations. The emer-gence of a dense [log nH(cm−3) & 7] new outflowabsorption-line system in PG 1411+442 was reportedin Hamann et al. (2019b) and the detailed inferrence ofa distance . 0.4 pc from the quasar is not repeated here.

We present the archival COS spectra for the other threeobjects in Figure 11, normalized to the same FUV con-tinuum level to emphasize absorption profile variations.In PG 1001+054 (Fig. 11a), the dramatic (72%) de-

crease in the FUV continuum emission from June 2014to March 2019 is accompanied by a strengthening of thebroad Lyα, N V, and O VI emission lines in terms ofequivalent widths but no obvious change in the mini-BAL profiles. In PG 1126−041 (Fig. 11b), the moremodest (20%) decrease of the continuum emission fromApril 2012 to June 2014 are not accompanied by anyobvious variations in the equivalent widths of any of thebroad emission and absorption lines except for the mostblueshifted N V absorption features below −3000 km s−1

which show variations on timescales perhaps as shortas 12 days. The broad emission and absorption linesin PG 2214+139 (Fig. 11c) show no variations betweenNovember 2011 and September 2021 despite a 26% in-crease in the strength of the FUV continuum emission.The fast 12-day variability of the high-velocity N V

mini-BAL in PG 1126−041 may be interpreted in twodifferent ways. One possibility is that transverse mo-tions of the outflowing clouds across our line of sightto the continuum source and BELR are responsiblefor these changes (as in PG 1411+442; Hamann et al.2019b). A variant on this idea is that the changes inprofiles are due instead to the dissolution and creationof the absorbing clouds/clumps in the outflow as theytransit in front of the continuum source. In the otherscenario, changes in the ionization structure of the ab-sorbing clouds due to changes in the incident quasar fluxcause the absorbing N V and O VI columns to vary andreproduce the observations. If this is the case, the vari-ability timescale sets a constraint on the ionization or re-combination timescale, which depends solely on the inci-dent ionizing continuum and gas density (∼ 105 yrs/nH ,where nH is the number density of the clouds in cm−3;e.g., He et al. 2019).This last scenario predicts that changes in the FUV

continuum of the quasar will produce changes in themini-BAL. While changes are indeed observed in boththe FUV continuum emission and high-velocity N Vmini-BAL of PG 1126−041, the amplitudes of thesechanges are not correlated. From 2012-04-15 to 2014-06-01, the continuum emission strengthened while the N Vmini-BAL weakened. From 2014-06-01 to 2014-06-12,both the continuum emission and N V mini-BAL weak-ened. Finally, from 2014-06-12 to 2014-06-28, the con-tinuum emission remained constant to within 1% but theN V mini-BAL strengthened slightly. This lack of a di-rect connection between variations in the continuum andthe N V mini-BAL seems to disfavor the scenario where

Highly Ionized N V and O VI Outflows in the QUEST Quasars 21

(b) PG 1126-041(a) PG 1001+054

(c) PG 2214+139

Figure 11. Multi-epoch comparisons of the mini-BALs in (a) PG 1001+054, (b) PG 1126−041, and (c) PG 2214+139. Allspectra are normalized to match the continuum level blueward of Lyα or redward of N V and O VI. The multiplicative factoris indicated in the caption.

the mini-BAL variations are associated with changes inthe ionization structure of the absorbing clouds, unlesslog nH(cm−3) . 5-6 in which case r >> rout and timedelays associated with the finite recombination timescalecould be at play (cf. Hamann et al. 2019b).While more detailed modeling of the mini-BAL of

PG 1126−041 is beyond the scope of the present paper,the fact that the mini-BAL variations are only observedin N V and only at high velocities may be an opticaldepth effect: the cloud complex that produces the Lyαand O VI mini-BALs and low-velocity N V mini-BALmay be so optically thick to be immune to variations inthe ionizing continuum or tangential movement of the

absorbing gas across the continuum source (we returnto this topic in Sec. 7.4 below).

7.3. Driving Mechanisms of the Mini-BALs

As reviewed in, for instance, Veilleux et al. (2020),the absorbing clouds making up the mini-BALs may bematerial (1) entrained in a hot, fast-moving fluid, or (2)pushed outward by radiation or cosmic ray pressure, or(3) created in-situ from the hot wind material itself. Inthe first two scenarios, the equation of motion of theoutflowing absorbers of massMabs that subtends a solidangle Ωabs is

d

dt[Mabs(r) r] = Ωabs r

2 (Pth + PCR + Pjet)

22 Veilleux et al.

+

(Ωabs4π

)(τLbol

c

)− GM(r)Mabs(r)

r2, (8)

where M(r) is the galaxy mass enclosed within a radiusr and τ is a volume- and frequency-integrated opticaldepth that takes into account both single- and multiple-scattering processes in cases of highly optically thickclouds (Hopkins et al. 2014, 2020).2 The terms on theright in Equation 8 are the forces due to the thermal,cosmic ray, and jet ram pressures, the radiation pressure,and gravity, respectively. Magneto-hydrodynamical ef-fects are assumed to be negligible at the distances ofthese absorbing clouds. The quasars in our sample donot have powerful radio jets so the jet ram pressure termcan safely be neglected. Similarly, the relatively mod-est radio luminosities of the mini-BAL quasars relativeto their optical and bolometric luminosities (Column 6in Table 1) suggest that cosmic-ray electrons do notplay an important dynamical role in accelerating theBAL clouds. Indeed the fraction of BAL quasars seemsto vary inversely with the radio loudness parameter, R(Column 5 in Table 1; e.g., Becker et al. 2001; Shankaret al. 2008). Below, we consider the remaining thermaland radiation pressure terms separately. In reality, thesepressure forces may act together to drive the mini-BALoutflows (see Sec. 7.4 for a closer look at the mini-BALPG 1126−041 in this context).

7.3.1. Thermal Wind and Blast Wave

For many years, ram-pressure acceleration of pre-existing clouds has been considered a serious contenderto explain BALs in quasars given the need for a muchhotter, rarefied medium to confine the clouds as they arebeing accelerated (e.g. Weymann et al. 1985). However,it is notoriouly difficult to accelerate dense gas cloudsfrom rest up to the typical (mini-)BAL velocities by awarm, fast thermal wind without destroying them inthe process through Rayleigh-Taylor fragmentation andshear-driven Kelvin-Helmholtz instabilities (e.g. Cooperet al. 2009; Scannapieco & Brüggen 2015; Schneider &Robertson 2015, 2017). Radiative cooling and magneticfields may act to slow down cloud disruption (Marcoliniet al. 2005; Cooper et al. 2009; Banda-Barragán et al.2016, 2018, 2020; Grønnow et al. 2018). Radiative cool-ing of the warm mixed gas can not only prevent disrup-tion, but it may cause the cloud to grow in mass (e.g.,Gronke & Oh 2018, 2020; Girichidis et al. 2021), al-

2 More explicitly, τ ≡ (1 − e−τsingle )(1 + τeff,IR). The value ofτ therefore ranges from ∼ τsingle = τUV/optical << 1 in theoptically thin case to ∼ (1 + τeff,IR) & 1 in the infrared optically-thick limit (the effective infrared optical depth, τeff,IR, is alsosometimes called the “boost factor”; Veilleux et al. (2020)).

though there are caveats (Schneider et al. 2020). An al-ternative scenario is that the BAL and mini-BAL cloudsare created in-situ via thermal instabilities and conden-sation from the hot gas with a cooling time shorterthan its dynamical time (Efstathiou 2000; Silich et al.2003). This is the idea behind the “blast wave” simula-tions of Richings & Faucher-Giguère (2018a,b); see alsoWeymann et al. (1985); Zubovas & King (2012, 2014);Faucher-Giguère & Quataert (2012); Nims et al. (2015).In these simulations, a fast (presumably X-ray emit-

ting) AGN wind with outward radial velocity of 30,000km s−1 is injected in the central 1 pc and collides vio-lently with the host ISM. The resulting shocked windmaterial reaches a very high temperature (∼1010 K;Nims et al. 2015) that does not efficienty cool, butinstead propagates outward as an adiabatic (energy-driven) hot bubble. This expanding bubble sweeps upgas and drives an outer shock into the host ISM raisingits temperature to a few × 107 K (Nims et al. 2015). Ra-diative cooling of the shocked ISM eventually becomesimportant and the outflowing material reforms as coolneutral and molecular gas, but by that time, the out-flowing material has acquired a significant fraction of theinitial kinetic energy of the hot wind. These simulationspredict that the cooling radius, i.e. the radius at whichthe gas cools to 104 K, increases from 100 pc to 1 kpcfor AGN with luminosities from 1044 to 1047 erg s−1, re-spectively (Fig. 7 of Richings & Faucher-Giguère 2018b).This cooling radius is also the expected location of thegas clouds producing the N V and O VI mini-BALs, asthe cooling gas rapidly transitions from ∼ 107 K to ∼104 K. This large radius is well outside the lower limit onthe distance of the mini-BAL from the quasars derivedabove so it is not inconsistent with our data. However,one should note that the inner X-ray wind in quasarsis presumed to much smaller in reality than the valueof 1 pc assumed in the simulations so the cooling ra-dius may have to be scaled down accordingly. Moreover,these models do not address how the BELRs would berestored after the passage of the blast wave. Finally,the detailed analysis of the BAL in PG 1411+442 firmlyrule out (at least in that case) absorption at these largedistances.

7.3.2. Radiative Acceleration

Although ram-pressure acceleration has been a seriouscontender, overall the favored explanation for the largevelocities of the BALs and mini-BALs is that the gas ab-sorbers have been accelerated by the radiation pressureforces associated with the intense radiation field thatis emanating from the quasars (e.g., Arav et al. 1994;Giustini & Proga 2019). Strong support for this sce-

Highly Ionized N V and O VI Outflows in the QUEST Quasars 23

nario comes from the observed trends for the maximumvelocity of the absorption to increase on average with in-creasing optical, UV, or bolometric luminosity and theEddington ratio (e.g., Perry & O’Dell 1978; Brandt et al.2000; Laor & Brandt 2002; Ganguly et al. 2007; Gan-guly & Brotherton 2008; Gibson et al. 2009b; Zhanget al. 2014). Note, however, that the overall correla-tions noted in these studies are often quite modest andsometimes only visible when considering the upper enve-lope of the velocity distribution and only when the sam-ple of AGN span 2-3 orders of magnitude in luminosity(sometimes combining NALs, mini-BALs, and BALs to-gether). While more recent studies (e.g., Rankine et al.2020) have confirmed and indeed strengthened the exis-tence of some of these correlations, all of the cited resultsrelate to the C IV absorption, rather than the N V andO VI features. The statistics on N V and particularlyO VI absorbers are much poorer.Far Ultraviolet Spectroscopic Explorer (FUSE) obser-

vations of Seyfert galaxies of relatively low luminosities(1038 − 1042 erg s−1) show either no or very weak trendsof increasing maximum velocities with increasing lumi-nosities and no trend at all with the Eddington ratio(Kriss 2004a,b; Dunn et al. 2008). O VI and N V BALsin high-redshift, high-luminosity quasars (Baskin et al.2013; Hamann et al. 2019a) have maximum velocitiesthat correlate with their C IV counterparts, but therange in AGN luminosity of their sample is too small totest the luminosity dependence of the maximum veloci-ties. More fundamentally, there is also a trend betweenline widths and optical depth. The most extreme exam-ple of this trend is P V, which coexists with C IV havingthe same ionization requirements, but is always weakerand narrower than C IV (Hamann et al. 2019a). Thereason is that P V traces only the highest column den-sity regions with smaller covering fractions, while C IVcan have significant absorption in more diffuse gas occu-pying a larger volume. This evidence for optical depth-dependent covering factors is a signature of inhomoge-neous partial covering.Overall, given the complex results from these previous

studies, it is perhaps not surprising to find no significantcorrelations in our sample of QUEST quasars between(maximum) outflow velocities and the AGN luminosities(Sec. 6.4). Theoretically, the noise in the trends betweenthe outflow kinematics and AGN luminosity is expectedin the radiative acceleration scenario given projectioneffects that reduce the measured outflow velocities andvariance in both the (minimum) launching radius (e.g.Laor & Brandt 2002) and efficiency of radiative acceler-ation associated with the complex micro-physics of thephoton interaction with the clouds - this complexity is

hidden in the quantity τ in equation 8. A similar trendof increasing variance in the maximum velocity with in-creasing AGN luminosity is observed in the other coolergas phases of AGN-driven outflows (e.g., Veilleux et al.2020; Fluetsch et al. 2020).Additional evidence that radiation pressure plays an

important role in accelerating the absorbers in quasarscomes from the significant dependence of the incidencerate, equivalent width, and weighted outflow velocitydispersion of the blueshifted absorbers on the X-rayproperties of the quasars. This effect has been reportedin numerous studies of nearby and distant AGN, basedlargely on C IV and Si IV (e.g., Laor & Brandt 2002;Gibson et al. 2009b), and is also clearly present in oursample of quasars based on N V and O VI (Sec. 6.1, 6.3,and 6.4). More specifically, we find that mini-BALs andBALs are broader, stronger, and more common amongX-ray faint quasars with steep optical-to-X-ray slopesαOX . −1.7 and hydrogen column densities NH in ex-cess of ∼ 1022 cm−2 (Sec. 6). This result is expectedin the context of radiative acceleration since the com-bined radiative force (“force multiplier”; Arav & Li 1994)is greatly reduced when the gas is over-ionized by thehard far-UV/X-rays, becoming too transparent to beradiatively accelerated. This over-ionized “failed-wind”material may act as a radiative shield to produce thespectral softening needed for efficient radiative accelera-tion of the outflow material downstream (Murray et al.1995; Proga & Kallman 2004; Proga 2007; Sim et al.2010). However, the strong near-UV absorption linesnear systemic velocity expected in this scenario are notobserved (Hamann et al. 2013). Alternatively, the spec-trum emerging from the accretion disk may be intrinsi-cally softer/fainter in the hard far-UV/X-rays than com-monly assumed (e.g. Laor & Davis 2014). Weak-lined“wind-dominated” quasars, such as Mrk 231, PHL 1811and its analogs, which are intrinsically faint and unab-sorbed in the X-rays, may be naturally explained in thisfashion (Richards et al. 2011; Wu et al. 2011; Luo et al.2015; Veilleux et al. 2016). While a connection shouldexist between the X-ray warm absorbers and the UVabsorption-line outflows, a direct one-to-one kinematiccorrespondence between the two classes of absorbers isoften not seen because the gas in the warm absorbersis too highly ionized to produce measurable lines in theUV spectra (Kaspi et al. 2000, 2001; Gabel et al. 2003;Kraemer et al. 2001; Krongold et al. 2003; Arav et al.2015; Laha et al. 2021, and references therein). We re-turn to this point in Section 7.4 when discussing themini-BAL in PG 1126−041 (the case of PG 1211+143is briefly discussed in Appendix A).

24 Veilleux et al.

Another observational characteristic of outflows thatfavors radiative acceleration is the phenomenon of line-locking observed in perhaps as many as ∼ 2/3 of allNAL and (mini-)BAL outflows (e.g., Hamann et al.2011; Bowler et al. 2014; Lu & Lin 2018; Mas-Ribas& Mauland 2019, and references therein). This is ob-served in outflows where multiple absorbers are presentbut are separated by the exact same velocity separa-tion as the doublet of C IV (499 km s−1), N V (962km s−1), or O VI (1650 km s−1). Previous studies haveshown that the probability of a line-locking signature ac-cidentally occurring over a relatively small redshift pathis negligible (e.g., Foltz et al. 1987; Srianand 2000; Sri-anand et al. 2002; Ganguly et al. 2003; Benn et al. 2005).Radiative acceleration is a natural explanation for line-locking (e.g., Mushotzky et al. 1972; Scargle 1973; Braun& Milgrom 1989).The best case for line-locking in our data is that of

PG 1351+640, where two deep absorption throughs aredetected in Lyα, extending over −[2200, 1500] km s−1

and −[1400, 600] km s−1, roughly separated by the ve-locity split of the N V doublet lines (∆v ≈ 900 − 1000km s−1 ≈ ∆vNV ; Fig. 12). This results in a N V mini-BAL that looks like a “triplet” in this object. Anothercase of line-locking may be present in PG 1126−041,where some of the deepest Lyα throughs are separatedby ∆v ≈ 900 − 1000 km s−1 ≈ ∆vNV (Fig. 3).Note, finally, that radiation pressure on dust grains

has also been invoked as an important contributor to theradiative acceleration given that (mini-)BAL QSOs, par-ticularly LoBALs and FeLoBALs, have more reddenedUV spectra than non-BAL QSOs (e.g., Allen et al. 2011;Hamann et al. 2019a, and references therein). This re-sult has been interpreted to mean that BAL and mini-BAL clouds have large columns of ionized + neutral gas(log NH(cm−2) & 23) and enough dust to provide ex-tinction equivalent to at least AV ∼ 1-2 mag. in somecases (this is discussed in Sec. 7.4 in the context ofPG 1126−041, but see also the results on Mrk 231 andPG 1411+442; Veilleux et al. 2016; Hamann et al. 2019b,respectively). Under these circumstances, the (mini-)BAL clouds may be subject to larger radiative forcesthan dustless clouds since the dust cross section, andthus τ in equation 8, in the UV optical is > 1-2 or-ders of magnitude than the Thompson scattering crosssection of electrons. Unfortunately, we do not have a re-liable reddening indicator of the FUV continuum emis-sion in our quasars, so we cannot directly compare ourdata with those of BAL and non-BAL QSOs. On theother hand, radiation that will be absorbed by the dustin the broad absorption line regions (BALRs) will bere-emitted in the infrared so our measurements of the

PG1351+640, z=0.088

−4000 −3000 −2000 −1000 0Velocity (km/s)

0.00.2

0.4

0.6

0.8

1.0

Nor

mal

ized

Fλ

NV 1238

0.00.2

0.4

0.6

0.8

1.0

NV 1242

0.00.2

0.4

0.6

0.8

1.0

Lyα

Figure 12. Line-locking in PG 1351+640. The two deepLyα absorption throughs are separated by the velocity splitof the N V doublet lines, resulting in a N V complex thatlooks like a “triplet”.

infrared excess in our quasars may serve as a surrogatefor the amount of dust in the BALRs. The lack of ob-vious correlation between BAL properties in our quasarsample and the mid-, far-, and total (1 − 1000 µm) in-frared excesses (e.g., Figs. 7 and 10) indicates one of twothings: (1) radiative acceleration on dust is not impor-tant in the BALRs of these objects or (2) the variousinfrared excesses are dominated by dust emission fromoutside of the BALR, e.g. dust in the host galaxy itself.

7.4. P V Mini-BAL in PG 1126−041 and OtherQuasars

In this last section, we take a closer look at the mini-BAL system in PG 1126−041. A mini-BAL extendingfrom −1000 to −5000 km s−1 was first reported in theN V, C IV, and S IV absorption lines of this objectby Wang & Wang (1999); Wang et al. (1999), based onthe analysis of old low-resolution spectra obtained withIUE and the Goddard High-Resolution Spectrograph(GHRS) on HST. Variable and much faster (∼16,500km s−1) X-ray absorption has also been detected in thisobject (Wang & Wang 1999; Wang et al. 1999; Gius-

Highly Ionized N V and O VI Outflows in the QUEST Quasars 25

tini et al. 2011). Interestingly, this object is among theleast luminous AGN (log LBOL/L = 11.52) in our sam-ple, intermediate between quasars and typical Seyfert 1galaxies. Mini-BALs with outflow velocities of up to∼ 5000 km s−1 and widths (FWHMrms ≡ 2.355 σrms)> 1000 km s−1 are relatively rare in such low-luminositysystems (e.g., Kriss 2004a,b; Dunn et al. 2008; Crenshaw& Kraemer 2012). On the other hand, PG 1126−041is also the object in our sample with the steepest X-ray to optical index (αOX = −2.13, a virtual tie withPG 1001+054, which also harbors a mini-BAL) andis among those with the largest infrared excess (Table1), reinforcing the view expressed in Section 7.3.2 thatX-ray absorbed or intrinsically weak quasars are morelikely to host BALs and mini-BALs.Apart from the line-locking signatures found in the

N V mini-BAL of this object, which we argued in Sec. 7.3favors radiative driving, the most remarkable aspect ofthis mini-BAL is the detection of a narrow P V λλ1117,1128 cloud at a velocity of −2200 km s−1 (Fig. 3). Largeionized-gas column densities are needed to produce thisfeature given the low abundance of phosphorus relativeto hydrogen (log(P/H) = −5.54; Lodders 2003). Themulti-component fit of each line in the P V doublet re-quires two components with nearly identifical medianvelocities (−2196 and −2203 km s−1) but different ve-locity dispersions (24 and 74 km s−1), covering factors(0.37 and 0.10, respectively), and optical depths (0.7and 2.6, respectively). The total equivalent width ofthis doublet is 0.3 Å.Taken at face value, the results from the fits suggest

that the P V lines are only moderately optically thickand therefore more reliable indicators of the total ion-ized column densities of this cloud than the highly sat-urated N V and O VI features. This is supported bythe ∼2:1 intensity ratio of the P V lines. An opticaldepth of order unity in P V λ1128 implies an ionizedhydrogen column density log NH(cm−2) ≈ 22.3, assum-ing a solar P/H abundance ratio and ionization correc-tions based on detailed photoionization calculations fortypical BALs and mini-BALs (ionization parameters logU & −0.5; Hamann 1998; Leighly et al. 2011; Borguetet al. 2012, 2013; Baskin et al. 2014; Capellupo et al.2017; Moravec et al. 2017; Hamann et al. 2019a, andreferences therein). This column density is remarkablyconsistent with the expectations from radiation pressureconfined cloud models (Baskin et al. 2014).This value of the total column density may be used to

estimate the minimum kinematic energy of this outflow-ing cloud using (eq. 2 from Hamann et al. 2019a)

Ekin = 1.5× 1052(

Q0.15

)(NH

2×1022 cm−2

)(r

1 pc

)2

×(

v2200 km s−1

)2ergs, (9)

where Q is an approximate global outflow covering fac-tor based on the incidence of mini-BALs in the SDSSquasars (Trump et al. 2006; Knigge et al. 2008; Gibsonet al. 2009b; Allen et al. 2011) and r = 1 pc is a place-holder radial distance that we adopt for illustration pur-poses (it may underestimate the actual distance of theabsorbers from the source; see Sec. 7.2 and Arav et al.2020, and references therein). Following Hamann et al.(2019b), we estimate the time-averaged kinetic energyluminosity, Lkin, by dividing Ekin by a characteristicflow time, r/v ≈ 450 yr. This yields Lkin & 1 × 1042

erg s−1. In these units, LBOL = 1.3 × 1045 erg s−1

so Lkin/LBOL & 0.001. Taken at face value, this ratiois too small to significantly affect the evolution of thegalaxy host (e.g., Lkin & 0.005 LEdd is needed accordingto Hopkins & Elvis 2010), unless (1) r is severely un-derestimated or (2) the other clouds at lower and highervelocities involved in this mini-BAL contribute signifi-cantly to Lkin. This second possibility seems unlikelygiven the lack of P V detection in these clouds whichsuggests column densities log NH(cm−2) . 22.Next, we use the total column density of the P V

cloud in PG 1126−041 to estimate the time-averagedmomentum outflow rate of this cloud, p = 2Lkin/v ≈1 × 1034 dynes, and compare this value with the radi-ation pressure, LBOL/c = 4 × 1034 dynes. Given thatp/(LBOL/c) ∼ 0.2, radiation pressure can thus in princi-ple accelerate this cloud, although a contribution from athermally driven wind as detailed in Section 7.3.1 cannotbe formally ruled out.Finally, we apply Equation 9 and calculate Lkin/LBOL

for the other mini-BALs in our sample with solid andtentative P V detections. For PG 1411+442, the onlyother mini-BAL in the sample with a definite P V detec-tion, we get Lkin/LBOL & 0.01 and p/(LBOL/c) ∼ 1, foran outflow velocity of −1800 km s−1 (Table 3), a totalcolumn density log NH(cm−2) & 23.4, and a BELR-likedistance . 0.4 pc from the central light source derivedby Hamann et al. (2019b) using several absorption linesand detailed photoionization simulations. This BALmay thus be sufficient to impact the host galaxy evo-lution. P V is also tentatively detected in PG 1001+054and PG 1004+130 at velocities of ∼ [−4000, −6,000]km s−1 (Fig. 13). In both cases, the equivalent widths ofP V 1117 and 1128 are very similar, implying saturationand log NH(cm−2) & 22.3. These numbers yield out-flows with kinetic-to-bolometric luminosity ratios thatare higher than that of PG 1126−041 but lower thanthat of PG 1411+442, and thus marginally sufficient toimpact the host galaxy evolution. Overall, the mini-BALs in the QUEST quasars are less powerful than the

26 Veilleux et al.

P V BALs detected in ze & 1.6 SDSS quasars (Moravecet al. 2017) but perhaps more common (4/33 ∼ 10%)than at high luminosities/redshifts (detection rate ofonly 3−6% among the ze & 2.6 BAL quasars in theSDSS-III BOSS quasar catalog Capellupo et al. 2017).

(a) PG 1001+054

(b) PG 1004+130

Figure 13. Tentative detection of P V in (a) PG 1001+054and (b) PG 1004+130. The spectra have been heavily binnedto emphasize the broad but shallow features. The deep andnarrow absorption lines in the velocity range −[8,000, 14,000]km s−1 in the P V 1117, 1128 panels of both PG 1001+054and PG 1004+130 are due to intervening MW ISM (Si II1259, 1260 + Fe II 1260 and C II 1334 + C II∗ 1335, respec-tively).

8. SUMMARY

As part I of a HST FUV spectroscopic study of theQUEST (Quasar/ULIRG Evolutionary Study) sampleof local quasars and ULIRGs, we have conducted a uni-form analysis of the COS spectra of 33 z . 0.3 Palomar-Green quasars. The main conclusions from our analysisare the following:

1. Highly ionized outflows traced by blueshifted N Vλλ1238, 1243 and O VI λλ1032, 1038 absorp-tion lines with equivalent widths larger than ∼20 mÅ are present in about 60% of the QUESTquasars. This detection rate is similar to thatof warm-ionized outflows traced by blueshiftedC IV λλ1548, 1550 absorption lines in localSeyfert galaxies and more distant, higher luminos-ity quasars.

2. The N V and O VI features in the QUEST quasarsspan a broad range of properties, both in termsof equivalent widths (from 20 mÅ to 25 Å) andkinematics (outflow velocities from a few × 100km s−1 up to ∼ 10,000 km s−1).

3. The rate of incidence and equivalent widths ofthe highly ionized outflows are higher among X-ray weak sources with X-ray to optical spectralindices αOX . −1.7 and X-ray column densitieslog NH(cm−2) & 22. The weighted outflow ve-locity dispersions are highest in the X-ray weak-est sources with X-ray to optical spectral indicesαOX . −2. These results are qualitatively similarto AGN-driven warm ionized outflows traced bythe C IV λλ1548, 1550 absorption lines. These re-sults favor radiative acceleration of the absorbers,where the X-rays are either absorbed or intrinsi-cally weak in the wind-dominated systems. Line-locking is detected in the Lyα absorption throughsof one or two objects, providing additional evi-dence that radiation pressure plays an importantrole in accelerating these absorbers.

4. There is no significant trend between the weightedaverage velocity of the highly ionized outflows andthe properties of the quasars and host galaxies.This negative result is likely due in part to thefact that the range of properties of the QUESTquasar sample is narrow in comparison to those ofother studies.

5. Blueshifted P V broad absorption lines are clearlydetected in PG 1126−041 and PG 1411+442 (pre-viously reported in Hamann et al. 2019b), and also

Highly Ionized N V and O VI Outflows in the QUEST Quasars 27

posssibly in PG 1001+054 and PG 1004+130. Us-ing the results from the analysis of Hamann et al.(2019b), these features imply column densities of∼ 1022.3 cm−2 or larger and time-averaged outflowkinetic power to bolometric luminosity ratios of &0.1% if a conservatively small radial distance of 1pc from the P V absorbers is assumed.

Paper II of this series (Liu et al. 2021, in prep.) willpresent the results from our analysis of the COS spectraon the QUEST ULIRGs. These results will be combinedwith those from the present paper to provide a morecomplete picture of the gaseous environments of quasarsand ULIRGs as a function of host galaxy properties andage across the merger sequence from ULIRGs to quasars.

ACKNOWLEDGMENTS

The authors thank the anonymous referee for sugges-tions which improved this paper. SV, WL, and TMTacknowledge partial support for this work provided byNASA through grant numbers HST GO-1256901A andGO-1256901B, GO-13460.001-A and GO-13460.001-B,and GO-15662.001-A and GO-15662.001-B from theSpace Telescope Science Institute, which is operatedby AURA, Inc., under NASA contract NAS 5-26555.Based on observations made with the NASA/ESA Hub-ble Space Telescope, and obtained from the HubbleLegacy Archive, which is a collaboration between theSpace Telescope Science Institute (STScI/NASA), theSpace Telescope European Coordinating Facility (ST-ECF/ESA) and the Canadian Astronomy Data Cen-tre (CADC/NRC/CSA). The authors also made use ofNASA’s Astrophysics Data System Abstract Service andthe NASA/IPAC Extragalactic Database (NED), whichis operated by the Jet Propulsion Laboratory, Califor-nia Institute of Technology, under contract with the Na-tional Aeronautics and Space Administration.

Software: COSQUEST (Rupke 2021a), IFSFIT(Rupke 2014; Rupke & Veilleux 2015; Rupke 2021b),LINMIX_ERR (Kelly 2007), pymccorrelation (Privonet al. 2020), scipy (Virtanen et al. 2020), DRTOOLS(Rupke 2021c)

28 Veilleux et al.

APPENDIX

A. DETAILED RESULTS FROM THE SPECTRAL ANALYSIS

Figures 14a−14t present the fits to the detected N V, O VI, and P V absorption systems in our sample. The resultsfrom these fits are listed in Table 3 in the main body of the paper. Here we summarize the results from our spectralanalysis for each object in the sample.

PG 0007+106.—There are no associated N V absorbers in this system, although two deep blueshifted and redshiftedLyα absorption features are present at |v| < 400 km s−1.PG 0026+129.—There are no associated N V or O VI absorbers in this object.PG 0050+124 (I Zw 1).—N V and Lyα absorbers are detected at −553, −1315, and −1467 km s−1 in this object.

Variable warm absorbers at −1870 and −2500 km s−1 have been reported by Silva et al. (2018) in XMM-Newton RGSspectra obtained in 2015.PG 0157+001 (Mrk 1014).—There are no associated N V or O VI absorbers in this system.PG 0804+761.—This is a rare case for infall where a strong redshifted absorption system at +600 km s−1 is observed

in Lyα, and a corresponding weaker feature is also detected in N V and O VI. The stronger features in the panel ofFigure 14b labeled O VI 1037 are Fe II features from the MW ISM.PG 0838+770.—A weak low-|v| absorption feature is seen in Lyα but not N V or O VI.PG 0844+349.—A strong slightly redshifted double-component absorption feature, likely associated with tidal debris,

is present at +140 km s−1 and +190 km s−1 in both Lyα and N V, but the feature at ∼ 1288.9 Å has no correspondingN V and is presumed to be Lyα from intervening CGM.PG 0923+201.—A single broad absorber is observed at −3300 km s−1 in Lyα, Lyβ, and O VI 1032 and 1038 although

the glare of the geocoronal Lyα airglow truncates the blue wing of the O VI 1032 absorption line.PG 0953+414.—Two faint O VI features are detected at −140 and −1074 km s−1. The first feature is also visible

in Lyα and Lyβ, but the more blueshifted feature is not. The strong feature near 1271 Å cannot be fit with O VI andthus is likely Lyα from intervening CGM.PG 1001+054.—The multi-epoch COS spectra shown in Figure 11 are co-added for this analysis given the lack of

variability in the absorption line profiles (Sec. 7.2). This is a special case because of the way the broad, high-velocityN V absorption absorbs the Lyα profile and the O VI absorption is deep, nearly dark, and highly saturated. The threemethods discussed in Section 5 were attempted to deal with the Lyα + N V blend, and in the end method #3 was usedfor the final fit: (1) The use of polynomials/splines to fit the blue and red sides of Lyα is problematic because it doesnot appear to yield a symmetric Lyα line and seems to miss N V absorption that appears in Lyα. (2) A Lorentzianfit to Lyα works reasonably well in that it yields a symmetric line, but the line properties are highly unconstrained(particularly in terms of height) and sensitive to the choice of which continuum regions are fit. (3) After trying theQSO templates from Vanden Berk et al. (2001), Stevans et al. (2014), and Harris et al. (2016), we settled on the lastone. We scaled the template using a constant offset, a constant multiplier, and a scaling according to λp (arbitrarypower p) to account for differences between the spectral index and that of the template (following Harris et al. 2016).We fit only the blue side of Lyα (and the far-red side) as well as some continuum regions in between O VI and LyαN Vthat are fairly line-free. This underpredicts the strength of highly-ionized lines, but is the best compromise solution.If the lines are fit as well, the blue side of Lyα is not properly fit. The resulting fit to N V seems to work reasonablywell, although the fit results are obviously illustrative in terms of velocity space and certainly do not get the opticaldepth correct. There seems to be a narrow Lyα near −4000 km s−1 that is barely detected in N V, but the fit hasdifficulties capturing it. The broad, highly saturated, nearly dark O VI is fit over −[6000, 4000] km s−1 but not −[8000,6000] km s−1 to account for Lyβ contamination. Broad blueshifted P V 1117, 1128 at ∼ −[6000, 4000] km s−1 is alsotentatively detected in this object but no attempt is made to fit this faint feature (Fig. 13a).PG 1004+130.—This is another special case (see Wills et al. 1999; Brandt et al. 2000, for some previous analyses).

It is difficult to fit the “continuum” shape and O VI profile simultaneously. The present fit is the best availablecompromise. It is clear that Lyβ / O VI are interacting with each other so the O VI fit should be taken only asillustrative. There are higher-order Lyman lines in the spectrum that are not considered. The Lyγ region is shown butis not considered in the fits to O VI (the many narrow features in the Lyγ region are Galactic ISM lines). BlueshiftedP V 1117 + 1128 at −[6000, 4000] km s−1 is tentatively detected but no attempt is made to fit this faint feature (Fig.

Highly Ionized N V and O VI Outflows in the QUEST Quasars 29

13b).PG 1116+215.—There are several narrow absorption features blueward of Lyα, O VI, and Lyβ in this system, but

only two of them are paired in O VI λ1032 and λ1038, corresponding to −771 and −2825 km s−1. Presumably, theorphan features are intervening Lyα absorbers. Savage et al. (2014) also report an O VI system at 1175 and 1181 Å,but it is not shown here as it is very narrow and very high velocity, and is intervening CGM (there are nearby galaxiesat the same redshift Tripp et al. 1998).PG 1126−041.—Several separate COS/G130M spectra exist for this object (Table 2). Our analysis uses the co-added

spectrum. Broad blueshifted absorption features are clearly detected in Lyα, N V, and O VI (Fig. 3). A narrow P Vfeature is observed at −2200 km s−1, indicative of large ionized column densities. This P V detection is discussed inmore detail in Section 7.4. 2014 and 2015 COS/G160M spectra centered around ∼ 1450 Å show strong C IV λλ1548,1550 absorption throughs similar to those of N V and O VI, but only weak and narrow blueshifted S IV λλ1392, 1402absorption features with |v| . 500 km s−1; the analysis of these features is beyond the scope of the present study.PG 1211+143.—There are no associated N V absorption systems in this object. A broad absorption feature at the

observed wavelength of 1240 Å has been interpreted by Kriss et al. (2018) as a highly blueshifted (−16,980 km s−1)and broad (FWHM ≈ 1080 km s−1) Lyα absorption from a fast wind. It may correspond to one of the ultra-fastoutflowing systems detected in the X-rays (Danehkar et al. 2018, and references therein). However, since it is notdetected in N V it is not included in the statistical analysis of Section 6.PG 1226+023 (3C 273).—The FUV spectrum of 3C 273 has been used extensively to study the low-z IGM (e.g.,

Tripp et al. 2008; Savage et al. 2014). There are no associated N V or O VI absorption systems in this object.PG 1229+204.—There are no associated N V absorption systems in this object. The many unidentified features

shortward of Lyα in the quasar rest-frame (e.g., 1289.5, 1282, 1223.2, 1220.4 Å) may be Lyα from intervening CGM.Mrk 231.—The 2011 COS/G130M spectrum of this object was presented in Veilleux et al. (2013a), while the

2014 COS/G140L and G230L spectra were presented in Veilleux et al. (2016). The COS/G130M spectrum shows Lyαemission that is broad (& 10,000 km s−1) and highly blueshifted (centroid at ∼ −3500 km s−1). In contrast, blueshiftedabsorption features are only present above ∼2200 Å. These results have been discussed in details in Veilleux et al.(2016), and this discussion is not repeated here. This outflow is considered a non-detection in our analysis in Section 6since it has no N V absorption systems, but has all of the characteristics of a FeLoBAL at visible and NUV wavelengthsand is considered as such in our discussion (Sec. 7).PG 1302−102.—There are no associated O VI absorbing systems in this object, but several intervening Lyα and

metal-line systems have been reported by Cooksey et al. (2008).PG 1307+085.—The weak O VI absorber at ∼ −3400 km s−1 is also detected in Lyα and Lyβ. The other feature

at ∼ −3600 km s−1 is seen only in Lyβ and Lyα and is presumed to be from intervening CGM.PG 1309+355.—A broad absorption feature is visible in Lyα, extending blueward to ∼ −1600 km s−1. This feature

is also detected in both O VI lines, but is truncated by the gap associated with the strong geocoronal Lyα emission.Strong absorption features are detected coincident with P V 1117 and 1128 near systemic velocity, but given theirequivalent widths they are likely of Galactic origin (e.g. C II 1334).PG 1351+640.—Two deep broad absorption throughs are detected in N V and Lyα of this object, extending over

[−2200, −1500] km s−1 and [−1400, −600] km s−1. They are roughly separated by the velocity separation of the N Vdoublet lines, resulting in a N V “triplet”, so this is a good case for line locking (Sec. 7.3.2). The template fit to the Lyαline emission works well in the very narrow region in which it is used, but mostly because it gets the smooth profile inthis region better than a spline+polynomial fit. The central peak is steeper than the template, and it is possible thatthe bluest absorption region in Lyα at −[3500, 2500] km s−1 is caused by a weak peak rather than a true absorption,since it does not line up with anything in N V. The N V template fit, again over this limited range, is quite good andmore trustworthy over a wider wavelength range. It is actually quite comparable to the spline fit. It does not get thesteep central peak in Lyα on the red side.PG 1411+442.—This object has been the subject of a detailed analysis by Hamann et al. (2019b). A deep broad

absorption through that extends over [−2800, −900] km s−1 is present in N V and Lyα. Strong P V absorption isalso detected over −[2200, 1400] km s−1 in this object. The absorption profiles presented here should be taken only asillustrative. They are produced using a simple spline/polynomical fit to the line emission + continuum. The quasartemplate is much too broad to match the observed emission line profiles of Lyα and N V.PG 1435−067.—There are no associated N V or O VI absorption systems in this object, but a faint Lyα absorber

appears to be present at ∼1369.3 Å or ∼ −700 km s−1. The other bluer narrow absorption features are likely Lyα

30 Veilleux et al.

absorption from the intervening CGM.PG 1440+356.—Two blueshifted N V absorbers are observed at −2190 and −1610 km s−1, and a fainter redshifted

one at +460 km s−1 is also detected in both lines of the doublet. Note that the strongest of the blueshifted N V systemsdoes not have a good match in velocity space with Lyα. The strong line near 1334 Å is produced in the Galactic ISM,while the strong line blueward of Lyα is presumably produced by intervening CGM. The Lyα line at +200 km s−1

may be systemic within the uncertainties on the redshift or a signature of inflow.PG 1448+273.—A broad (FWZI ≈ 700 km s−1) multi-component absorption feature is detected in both N V and

Lyα. The narrow Lyα feature at ∼ −2700 km s−1 is likely produced by intervening CGM, while the deep saturatedfeature at 1280 Å is C I from Galactic ISM.PG 1501+106.—Three strong absorption features are seen blueward of Lyα at ∼ 1253.0, 1255.8, and 1257.2 Å,

corresponding to ∼ −1500, −800, and −500 km s−1, respectively (the absorption near 0 km s−1 is at least partly dueto S II from Galactic ISM). Despite the chip gap at the position of the N V doublet, the present data allow us to ruleout the presence of a N V λ1238 counterpart to the most highly blueshifted of these Lyα features since it would lienear ∼1276.5 Å and is not present in the data.PG 1613+658.—Two weak narrow absorption features are detected at −3503 and −3764 km s−1 in N V, O VI,

and Lyα. The other features around N V are of Galactic ISM origin, while those in O VI and Lyα are likely due tointervening CGM.PG 1617+175.—A broad (∼1000 km s−1) blueshifted absorption feature centered around ∼ −3000 km s−1 is detected

in N V, O VI, and Lyα. Three narrow absorbers at −3300, −1630, and −1040 km s−1 are also detected in O VI andLyα but are very weak or absent in N V.PG 1626+554.—The two distinct blueshifted Lyα absorption features at 1378.8 Å and 1374.7 Å, corresponding to−740 and −570 km s−1 in the quasar rest frame, are also detected in Lyβ but not in N V. A faint depression at 1166.9Å may be the −570 km s−1 counterpart of O VI 1032 but it is not detected in the fainter O VI 1038. This feature wasjudged too uncertain to be a detection in our analysis.PG 2130+099.—Two deep and narrow absorption features at ∼ −1500 and 0 km s−1 are detected in N V and Lyα

of this object.PG 2214+139.—Our analysis is based on the co-added spectrum of this object given the lack of variability in the

absorption lines between 2011 and 2012 (Fig. 11). This object shows complex N V troughs that extend from ∼ −3400to −400 km s−1 and are loosely matched to the complex absorption feature at Lyα, except for the sharp Lyα absorptionfeature near −100 km s−1, which is not detected in N V.PG 2233+134.—This object shows a faint and narrow absorber at ∼ −200 km s−1 in O VI and Lyβ. The Lyβ

absorber at ∼ −1300 km s−1 also seems to have a weak O VI counterpart but the fit is inconclusive and is thereforenot included in the statistics for this object. The other stronger lines in this spectral region do not match GalacticISM features so they are likely produced by intervening CGM.PG 2349−014.—There are no N V or O V λ1032 absorbers in this object (O VI λ1038 is lost in the glare of geocoronal

Lyα).

Highly Ionized N V and O VI Outflows in the QUEST Quasars 31

(a)

(b)

(c)

(d

)

(e

)

(f)

(g)

(h)

(

j)

(i)

(k)

(l)

PG00

50+1

24, z

=0.0

59

ï200

0ï1

500

ï100

0ï5

000

500

1000

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG08

04+7

61, z

=0.1

00

020

040

060

080

010

00Ve

locit

y (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV 1

238

0.0

0.2

0.4

0.6

0.8

1.0

NV 1

242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG08

44+3

49, z

=0.0

64

ï400

ï200

020

040

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG09

23+2

01, z

=0.1

92

ï400

0ï3

000

ï200

0ï1

000

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG09

53+4

14, z

=0.2

34

ï150

0ï1

000

ï500

050

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

0.0

0.2

0.4

0.6

0.8

1.0

Lyb

PG10

01+0

54, z

=0.1

61

ï100

00ï8

000

ï600

0ï4

000

ï200

00

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

0.0

0.2

0.4

0.6

0.8

1.0

Lya

PG10

04+1

30, z

=0.2

41

ï1.2×1

04ï9

.0×1

03ï6

.0×1

03ï3

.0×1

030

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

0.0

0.2

0.4

0.6

0.8

1.0

Lya

PG11

16+2

15, z

=0.1

76

ï400

0ï3

000

ï200

0ï1

000

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG11

26ï0

41, z

=0.0

60

ï500

0ï4

000ï3

000ï2

000ï1

000

010

00Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

PV 1

117

0.0

0.2

0.4

0.6

0.8

1.0

PV 1

128

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG13

07+0

85, z

=0.1

55

ï450

0ï4

000

ï350

0ï3

000

ï250

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG13

09+3

55, z

=0.1

83

ï150

0ï1

000

ï500

050

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

0.0

0.2

0.4

0.6

0.8

1.0

Lya

PG13

51+6

40, z

=0.0

88

ï400

0ï3

000

ï200

0ï1

000

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

Fig

ure

14.[PartI]Interlinecompa

risonof

theab

sorbingsystem

sin

each

quasar

withde

tected

NV

orO

VIab

sorption

lines,p

rodu

cedby

plotting

theno

rmalized

spectrum

invelocity

spacerelative

tothequ

asar

rest

fram

e.The

data

arein

black,

theindividu

alcompo

nentsused

tofit

theab

sorption

profi

lesareshow

nin

cyan

,an

dtheoverallfit

isshow

nin

purple.The

velocity

centroidsof

themainab

sorbingsystem

sareindicatedby

vertical

red(L

yα,Lyβ

,N

V,P

V)an

dblue

(OVI)

dotted

lines.

32 Veilleux et al.

(m)

(n)

(o)

(p)

(q

)

(

r)

(s)

(t)

PG14

11+4

42, z

=0.0

90

ï500

0ï4

000

ï300

0ï2

000

ï100

00

1000

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

PV 1

117

0.0

0.2

0.4

0.6

0.8

1.0

PV 1

128

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG14

40+3

56, z

=0.0

77

ï300

0ï2

000

ï100

00

1000

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG14

48+2

73, z

=0.0

65

ï600

ï400

ï200

020

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG16

13+6

58, z

=0.1

29

ï450

0ï4

000

ï350

0ï3

000

ï250

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG16

17+1

75, z

=0.1

14

ï500

0ï4

000

ï300

0ï2

000

ï100

00

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

PG21

30+0

99, z

=0.0

63

ï200

0ï1

500

ï100

0ï5

000

500

Velo

city

(km

/s)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG22

14+1

39, z

=0.0

66

ï400

0ï3

000

ï200

0ï1

000

010

00Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

NV

1238

0.0

0.2

0.4

0.6

0.8

1.0

NV

1242

0.0

0.2

0.4

0.6

0.8

1.0

Ly_

PG22

33+1

34, z

=0.3

27

ï500

ï400

ï300

ï200

ï100

0Ve

loci

ty (k

m/s

)

0.0

0.2

0.4

0.6

0.8

1.0

Normalized Fh

OVI

103

2

0.0

0.2

0.4

0.6

0.8

1.0

OVI

103

8

0.0

0.2

0.4

0.6

0.8

1.0

Ly`

Fig

ure

13.[PartII]S

ameas

previous

figure.

Highly Ionized N V and O VI Outflows in the QUEST Quasars 33

REFERENCES

Allen, J. T., Hewett, P. C., Maddox, N., Richards, G. T., &Belokurov, V. 2011, MNRAS, 410, 860,doi: 10.1111/j.1365-2966.2010.17489.x

Arav, N., Kaastra, J., Kriss, G. A., et al. 2005, ApJ, 620,665, doi: 10.1086/425560

Arav, N., & Li, Z.-Y. 1994, ApJ, 427, 700,doi: 10.1086/174177

Arav, N., Li, Z.-Y., & Begelman, M. C. 1994, ApJ, 432, 62,doi: 10.1086/174549

Arav, N., Moe, M., Costantini, E., et al. 2008, ApJ, 681,954, doi: 10.1086/588651

Arav, N., Xu, X., Miller, T., Kriss, G. A., & Plesha, R.2020, ApJS, 247, 37, doi: 10.3847/1538-4365/ab66af

Arav, N., Chamberlain, C., Kriss, G. A., et al. 2015, A&A,577, A37, doi: 10.1051/0004-6361/201425302

Banda-Barragán, W. E., Brüggen, M., Federrath, C., et al.2020, MNRAS, 499, 2173, doi: 10.1093/mnras/staa2904

Banda-Barragán, W. E., Federrath, C., Crocker, R. M., &Bicknell, G. V. 2018, MNRAS, 473, 3454,doi: 10.1093/mnras/stx2541

Banda-Barragán, W. E., Parkin, E. R., Federrath, C.,Crocker, R. M., & Bicknell, G. V. 2016, MNRAS, 455,1309, doi: 10.1093/mnras/stv2405

Barvainis, R. 1987, ApJ, 320, 537, doi: 10.1086/165571Baskin, A., & Laor, A. 2005, MNRAS, 356, 1029,doi: 10.1111/j.1365-2966.2004.08525.x

—. 2018, MNRAS, 474, 1970, doi: 10.1093/mnras/stx2850Baskin, A., Laor, A., & Hamann, F. 2013, MNRAS, 432,1525, doi: 10.1093/mnras/stt582

Baskin, A., Laor, A., & Stern, J. 2014, MNRAS, 445, 3025,doi: 10.1093/mnras/stu1732

Becker, R. H., White, R. L., Gregg, M. D., et al. 2001,ApJS, 135, 227, doi: 10.1086/321798

Benn, C. R., Carballo, R., Holt, J., et al. 2005, MNRAS,360, 1455, doi: 10.1111/j.1365-2966.2005.09143.x

Bentz, M. C., & Katz, S. 2015, PASP, 127, 67,doi: 10.1086/679601

Bentz, M. C., Denney, K. D., Grier, C. J., et al. 2013, ApJ,767, 149, doi: 10.1088/0004-637X/767/2/149

Borguet, B. C. J., Arav, N., Edmonds, D., Chamberlain,C., & Benn, C. 2013, ApJ, 762, 49,doi: 10.1088/0004-637X/762/1/49

Borguet, B. C. J., Edmonds, D., Arav, N., Benn, C., &Chamberlain, C. 2012, ApJ, 758, 69,doi: 10.1088/0004-637X/758/1/69

Boroson, T. A., & Green, R. F. 1992, ApJS, 80, 109,doi: 10.1086/191661

Bowler, R. A. A., Hewett, P. C., Allen, J. T., & Ferland,G. J. 2014, MNRAS, 445, 359,doi: 10.1093/mnras/stu1730

Brandt, W. N., Laor, A., & Wills, B. J. 2000, ApJ, 528,637, doi: 10.1086/308207

Braun, E., & Milgrom, M. 1989, ApJ, 342, 100,doi: 10.1086/167578

Cameron, E. 2011, PASA, 28, 128, doi: 10.1071/AS10046Capellupo, D. M., Hamann, F., Shields, J. C., RodríguezHidalgo, P., & Barlow, T. A. 2012, MNRAS, 422, 3249,doi: 10.1111/j.1365-2966.2012.20846.x

Capellupo, D. M., Hamann, F., Herbst, H., et al. 2017,MNRAS, 469, 323, doi: 10.1093/mnras/stx870

Carilli, C. L., Wrobel, J. M., & Ulvestad, J. S. 1998, AJ,115, 928, doi: 10.1086/300253

Cicone, C., Maiolino, R., Sturm, E., et al. 2014, A&A, 562,A21, doi: 10.1051/0004-6361/201322464

Cooksey, K. L., Prochaska, J. X., Chen, H.-W., Mulchaey,J. S., & Weiner, B. J. 2008, ApJ, 676, 262,doi: 10.1086/528704

Cooper, J. L., Bicknell, G. V., Sutherland, R. S., &Bland-Hawthorn, J. 2009, ApJ, 703, 330,doi: 10.1088/0004-637X/703/1/330

Crenshaw, D. M., & Kraemer, S. B. 2012, ApJ, 753, 75,doi: 10.1088/0004-637X/753/1/75

Crenshaw, D. M., Kraemer, S. B., Boggess, A., et al. 1999,ApJ, 516, 750, doi: 10.1086/307144

Curran, P. A. 2014, arXiv e-prints, arXiv:1411.3816.https://arxiv.org/abs/1411.3816

Danehkar, A., Nowak, M. A., Lee, J. C., et al. 2018, ApJ,853, 165, doi: 10.3847/1538-4357/aaa427

Danforth, C. W., Keeney, B. A., Stocke, J. T., Shull, J. M.,& Yao, Y. 2010, ApJ, 720, 976,doi: 10.1088/0004-637X/720/1/976

Danforth, C. W., Keeney, B. A., Tilton, E. M., et al. 2016,ApJ, 817, 111, doi: 10.3847/0004-637X/817/2/111

Dasyra, K. M., Tacconi, L. J., Davies, R. I., et al. 2007,ApJ, 657, 102, doi: 10.1086/510552

de Vaucouleurs, G., de Vaucouleurs, A., Corwin, Herold G.,J., et al. 1991, Third Reference Catalogue of BrightGalaxies

Dunn, J. P., Crenshaw, D. M., Kraemer, S. B., & Trippe,M. L. 2008, AJ, 136, 1201,doi: 10.1088/0004-6256/136/3/1201

Efstathiou, G. 2000, MNRAS, 317, 697,doi: 10.1046/j.1365-8711.2000.03665.x

Elvis, M. 2000, ApJ, 545, 63, doi: 10.1086/317778Evans, A. S., Solomon, P. M., Tacconi, L. J., Vavilkin, T.,& Downes, D. 2006, AJ, 132, 2398, doi: 10.1086/508416

34 Veilleux et al.

Faucher-Giguère, C.-A., & Quataert, E. 2012, MNRAS,425, 605, doi: 10.1111/j.1365-2966.2012.21512.x

Filiz Ak, N., Brandt, W. N., Hall, P. B., et al. 2012, ApJ,757, 114, doi: 10.1088/0004-637X/757/2/114

—. 2013, ApJ, 777, 168, doi: 10.1088/0004-637X/777/2/168Fitzpatrick, E. L. 1999, PASP, 111, 63, doi: 10.1086/316293Fluetsch, A., Maiolino, R., Carniani, S., et al. 2019,MNRAS, 483, 4586, doi: 10.1093/mnras/sty3449

—. 2020, arXiv e-prints, arXiv:2006.13232.https://arxiv.org/abs/2006.13232

Foltz, C. B., Weymann, R. J., Morris, S. L., & Turnshek,D. A. 1987, ApJ, 317, 450, doi: 10.1086/165290

Gabel, J. R., Crenshaw, D. M., Kraemer, S. B., et al. 2003,ApJ, 583, 178, doi: 10.1086/345096

Ganguly, R., Bond, N. A., Charlton, J. C., et al. 2001, ApJ,549, 133, doi: 10.1086/319082

Ganguly, R., & Brotherton, M. S. 2008, ApJ, 672, 102,doi: 10.1086/524106

Ganguly, R., Brotherton, M. S., Cales, S., et al. 2007, ApJ,665, 990, doi: 10.1086/519759

Ganguly, R., Masiero, J., Charlton, J. C., & Sembach,K. R. 2003, ApJ, 598, 922, doi: 10.1086/379057

George, I. M., Turner, T. J., Netzer, H., et al. 1998, ApJS,114, 73, doi: 10.1086/313067

Gibson, R. R., Brandt, W. N., Gallagher, S. C., Hewett,P. C., & Schneider, D. P. 2010, ApJ, 713, 220,doi: 10.1088/0004-637X/713/1/220

Gibson, R. R., Brandt, W. N., Gallagher, S. C., &Schneider, D. P. 2009a, ApJ, 696, 924,doi: 10.1088/0004-637X/696/1/924

Gibson, R. R., Brandt, W. N., Schneider, D. P., &Gallagher, S. C. 2008, ApJ, 675, 985, doi: 10.1086/527462

Gibson, R. R., Jiang, L., Brandt, W. N., et al. 2009b, ApJ,692, 758, doi: 10.1088/0004-637X/692/1/758

Girichidis, P., Naab, T., Walch, S., & Berlok, T. 2021,MNRAS, doi: 10.1093/mnras/stab1203

Giustini, M., & Proga, D. 2019, A&A, 630, A94,doi: 10.1051/0004-6361/201833810

Giustini, M., Cappi, M., Chartas, G., et al. 2011, A&A,536, A49, doi: 10.1051/0004-6361/201117732

Gravity Collaboration, Sturm, E., Dexter, J., et al. 2018,Nature, 563, 657, doi: 10.1038/s41586-018-0731-9

Grier, C. J., Hall, P. B., Brandt, W. N., et al. 2015, ApJ,806, 111, doi: 10.1088/0004-637X/806/1/111

Gronke, M., & Oh, S. P. 2018, MNRAS, 480, L111,doi: 10.1093/mnrasl/sly131

—. 2020, MNRAS, 492, 1970, doi: 10.1093/mnras/stz3332Grønnow, A., Tepper-García, T., & Bland-Hawthorn, J.2018, ApJ, 865, 64, doi: 10.3847/1538-4357/aada0e

Guyon, O., Sanders, D. B., & Stockton, A. 2006, ApJS,166, 89, doi: 10.1086/505030

Haislmaier, K. J., Tripp, T. M., Katz, N., et al. 2021,MNRAS, 502, 4993, doi: 10.1093/mnras/staa3544

Hamann, F. 1998, ApJ, 500, 798, doi: 10.1086/305776Hamann, F., Barlow, T. A., Junkkarinen, V., & Burbidge,E. M. 1997a, ApJ, 478, 80, doi: 10.1086/303781

Hamann, F., Beaver, E. A., Cohen, R. D., et al. 1997b,ApJ, 488, 155, doi: 10.1086/304698

Hamann, F., Chartas, G., McGraw, S., et al. 2013,MNRAS, 435, 133, doi: 10.1093/mnras/stt1231

Hamann, F., Herbst, H., Paris, I., & Capellupo, D. 2019a,MNRAS, 483, 1808, doi: 10.1093/mnras/sty2900

Hamann, F., Kanekar, N., Prochaska, J. X., et al. 2011,MNRAS, 410, 1957,doi: 10.1111/j.1365-2966.2010.17575.x

Hamann, F., Kaplan, K. F., Rodríguez Hidalgo, P.,Prochaska, J. X., & Herbert-Fort, S. 2008, MNRAS, 391,L39, doi: 10.1111/j.1745-3933.2008.00554.x

Hamann, F., & Sabra, B. 2004, in Astronomical Society ofthe Pacific Conference Series, Vol. 311, AGN Physicswith the Sloan Digital Sky Survey, ed. G. T. Richards &P. B. Hall, 203. https://arxiv.org/abs/astro-ph/0310668

Hamann, F., Tripp, T. M., Rupke, D., & Veilleux, S. 2019b,MNRAS, 487, 5041, doi: 10.1093/mnras/stz1408

Hamilton, T. S., Casertano, S., & Turnshek, D. A. 2008,ApJ, 678, 22, doi: 10.1086/524182

Harris, D. W., Jensen, T. W., Suzuki, N., et al. 2016, AJ,151, 155, doi: 10.3847/0004-6256/151/6/155

He, Z., Wang, T., Liu, G., et al. 2019, Nature Astronomy,3, 265, doi: 10.1038/s41550-018-0669-8

Hewett, P. C., & Wild, V. 2010, MNRAS, 405, 2302,doi: 10.1111/j.1365-2966.2010.16648.x

Hickox, R. C., & Alexander, D. M. 2018, ARA&A, 56, 625,doi: 10.1146/annurev-astro-081817-051803

Ho, L. C., & Kim, M. 2009, ApJS, 184, 398,doi: 10.1088/0067-0049/184/2/398

Hopkins, P. F., & Elvis, M. 2010, MNRAS, 401, 7,doi: 10.1111/j.1365-2966.2009.15643.x

Hopkins, P. F., Grudić, M. Y., Wetzel, A., et al. 2020,MNRAS, 491, 3702, doi: 10.1093/mnras/stz3129

Hopkins, P. F., Kereš, D., Oñorbe, J., et al. 2014, MNRAS,445, 581, doi: 10.1093/mnras/stu1738

Hu, C., Li, S.-S., Guo, W.-J., et al. 2020, ApJ, 905, 75,doi: 10.3847/1538-4357/abc2da

Inoue, H., Terashima, Y., & Ho, L. C. 2007, ApJ, 662, 860,doi: 10.1086/517995

Ishibashi, W., Fabian, A. C., & Arakawa, N. 2021,MNRAS, 502, 3638, doi: 10.1093/mnras/stab266

Highly Ionized N V and O VI Outflows in the QUEST Quasars 35

Ishibashi, W., Fabian, A. C., & Maiolino, R. 2018,MNRAS, 476, 512, doi: 10.1093/mnras/sty236

Isobe, T., Feigelson, E. D., & Nelson, P. I. 1986, ApJ, 306,490, doi: 10.1086/164359

Jin, C., Ward, M., Done, C., & Gelbord, J. 2012, MNRAS,420, 1825, doi: 10.1111/j.1365-2966.2011.19805.x

Kaspi, S., Brandt, W. N., Maoz, D., et al. 2007, ApJ, 659,997, doi: 10.1086/512094

Kaspi, S., Brandt, W. N., Netzer, H., et al. 2000, ApJL,535, L17, doi: 10.1086/312697

Kaspi, S., Maoz, D., Netzer, H., et al. 2005, ApJ, 629, 61,doi: 10.1086/431275

Kaspi, S., Brandt, W. N., Netzer, H., et al. 2001, ApJ, 554,216, doi: 10.1086/321333

Kelly, B. C. 2007, ApJ, 665, 1489, doi: 10.1086/519947Kim, M., Ho, L. C., Peng, C. Y., Barth, A. J., & Im, M.2008, ApJS, 179, 283, doi: 10.1086/591796

Knigge, C., Scaringi, S., Goad, M. R., & Cottis, C. E. 2008,MNRAS, 386, 1426,doi: 10.1111/j.1365-2966.2008.13081.x

Kraemer, S. B., Crenshaw, D. M., & Gabel, J. R. 2001,ApJ, 557, 30, doi: 10.1086/321619

Kriss, G. A. 2004a, arXiv e-prints, astro.https://arxiv.org/abs/astro-ph/0411380

Kriss, G. A. 2004b, in The Interplay Among Black Holes,Stars and ISM in Galactic Nuclei, ed.T. Storchi-Bergmann, L. C. Ho, & H. R. Schmitt, Vol.222, 223–228, doi: 10.1017/S1743921304002157

Kriss, G. A., Lee, J. C., Danehkar, A., et al. 2018, ApJ,853, 166, doi: 10.3847/1538-4357/aaa42b

Krongold, Y., Nicastro, F., Brickhouse, N. S., et al. 2003,ApJ, 597, 832, doi: 10.1086/378639

Krug, H. B. 2013, PhD thesis, University of Maryland,College Park

Laha, S., Guainazzi, M., Piconcelli, E., et al. 2018, ApJ,868, 10, doi: 10.3847/1538-4357/aae390

Laha, S., Reynolds, C. S., Reeves, J., et al. 2021, NatureAstronomy, 5, 13, doi: 10.1038/s41550-020-01255-2

Lani, C., Netzer, H., & Lutz, D. 2017, MNRAS, 471, 59,doi: 10.1093/mnras/stx1374

Laor, A., & Brandt, W. N. 2002, ApJ, 569, 641,doi: 10.1086/339476

Laor, A., & Davis, S. W. 2014, MNRAS, 438, 3024,doi: 10.1093/mnras/stt2408

Leighly, K. M., Dietrich, M., & Barber, S. 2011, ApJ, 728,94, doi: 10.1088/0004-637X/728/2/94

Lodders, K. 2003, ApJ, 591, 1220, doi: 10.1086/375492Lu, W.-J., & Lin, Y.-R. 2018, ApJ, 863, 186,doi: 10.3847/1538-4357/aad411

Luo, B., Brandt, W. N., Hall, P. B., et al. 2015, ApJ, 805,122, doi: 10.1088/0004-637X/805/2/122

Lutz, D., Sturm, E., Janssen, A., et al. 2020, A&A, 633,A134, doi: 10.1051/0004-6361/201936803

Lyu, J., Rieke, G. H., & Shi, Y. 2017, ApJ, 835, 257,doi: 10.3847/1538-4357/835/2/257

Marcolini, A., Strickland, D. K., D’Ercole, A., Heckman,T. M., & Hoopes, C. G. 2005, MNRAS, 362, 626,doi: 10.1111/j.1365-2966.2005.09343.x

Martin, C. L., Dijkstra, M., Henry, A., et al. 2015, ApJ,803, 6, doi: 10.1088/0004-637X/803/1/6

Marziani, P., Sulentic, J. W., Dultzin-Hacyan, D., Calvani,M., & Moles, M. 1996, ApJS, 104, 37,doi: 10.1086/192291

Mas-Ribas, L., & Mauland, R. 2019, ApJ, 886, 151,doi: 10.3847/1538-4357/ab4efd

Moravec, E. A., Hamann, F., Capellupo, D. M., et al. 2017,MNRAS, 468, 4539, doi: 10.1093/mnras/stx775

Murray, N., Chiang, J., Grossman, S. A., & Voit, G. M.1995, ApJ, 451, 498, doi: 10.1086/176238

Mushotzky, R. F., Solomon, P. M., & Strittmatter, P. A.1972, ApJ, 174, 7, doi: 10.1086/151463

Nelson, D., Pillepich, A., Springel, V., et al. 2019, MNRAS,490, 3234, doi: 10.1093/mnras/stz2306

Nelson, E. J., Tacchella, S., Diemer, B., et al. 2021, arXive-prints, arXiv:2101.12212.https://arxiv.org/abs/2101.12212

Netzer, H., & Laor, A. 1993, ApJL, 404, L51,doi: 10.1086/186741

Netzer, H., Lutz, D., Schweitzer, M., et al. 2007, ApJ, 666,806, doi: 10.1086/520716

Nims, J., Quataert, E., & Faucher-Giguère, C.-A. 2015,MNRAS, 447, 3612, doi: 10.1093/mnras/stu2648

Onken, C. A., Ferrarese, L., Merritt, D., et al. 2004, ApJ,615, 645, doi: 10.1086/424655

Oppenheimer, B. D., Davies, J. J., Crain, R. A., et al. 2020,MNRAS, 491, 2939, doi: 10.1093/mnras/stz3124

Perry, J. J., & O’Dell, S. L. 1978, A&A, 62, 229Piconcelli, E., Jimenez-Bailón, E., Guainazzi, M., et al.2005, A&A, 432, 15, doi: 10.1051/0004-6361:20041621

Privon, G. C., Ricci, C., Aalto, S., et al. 2020, ApJ, 893,149, doi: 10.3847/1538-4357/ab8015

Prochaska, J. X., Wolfe, A. M., Tytler, D., et al. 2001,ApJS, 137, 21, doi: 10.1086/322542

Proga, D. 2007, ApJ, 661, 693, doi: 10.1086/515389Proga, D., & Kallman, T. R. 2004, ApJ, 616, 688,doi: 10.1086/425117

Rankine, A. L., Hewett, P. C., Banerji, M., & Richards,G. T. 2020, MNRAS, 492, 4553,doi: 10.1093/mnras/staa130

36 Veilleux et al.

Revalski, M., Crenshaw, D. M., Kraemer, S. B., et al. 2018,ApJ, 856, 46, doi: 10.3847/1538-4357/aab107

Reynolds, C. S. 1997, MNRAS, 286, 513,doi: 10.1093/mnras/286.3.513

Ricci, C., Trakhtenbrot, B., Koss, M. J., et al. 2017, ApJS,233, 17, doi: 10.3847/1538-4365/aa96ad

Richards, G. T., Kruczek, N. E., Gallagher, S. C., et al.2011, AJ, 141, 167, doi: 10.1088/0004-6256/141/5/167

Richings, A. J., & Faucher-Giguère, C.-A. 2018a, MNRAS,474, 3673, doi: 10.1093/mnras/stx3014

—. 2018b, MNRAS, 478, 3100, doi: 10.1093/mnras/sty1285Richings, A. J., Faucher-Giguère, C.-A., & Stern, J. 2021,MNRAS, 503, 1568, doi: 10.1093/mnras/stab556

Rupke, D. 2021a, doi: 10.5281/zenodo.5659382—. 2021b, doi: 10.5281/zenodo.5659520—. 2021c, doi: 10.5281/zenodo.5659554Rupke, D. S., Veilleux, S., & Sanders, D. B. 2005, ApJS,160, 87, doi: 10.1086/432886

Rupke, D. S. N. 2014, IFSFIT: Spectral Fitting for IntegralField Spectrographs. http://ascl.net/1409.005

Rupke, D. S. N., Gültekin, K., & Veilleux, S. 2017, ApJ,850, 40, doi: 10.3847/1538-4357/aa94d1

Rupke, D. S. N., & Veilleux, S. 2015, ApJ, 801, 126,doi: 10.1088/0004-637X/801/2/126

Sanders, D. B., Soifer, B. T., Elias, J. H., et al. 1988, ApJ,325, 74, doi: 10.1086/165983

Savage, B. D., Kim, T. S., Wakker, B. P., et al. 2014, ApJS,212, 8, doi: 10.1088/0067-0049/212/1/8

Scannapieco, E., & Brüggen, M. 2015, ApJ, 805, 158,doi: 10.1088/0004-637X/805/2/158

Scargle, J. D. 1973, ApJ, 179, 705, doi: 10.1086/151910Schlafly, E. F., & Finkbeiner, D. P. 2011, ApJ, 737, 103,doi: 10.1088/0004-637X/737/2/103

Schmidt, M., & Green, R. F. 1983, ApJ, 269, 352,doi: 10.1086/161048

Schneider, D. P., Richards, G. T., Hall, P. B., et al. 2010,AJ, 139, 2360, doi: 10.1088/0004-6256/139/6/2360

Schneider, E. E., Ostriker, E. C., Robertson, B. E., &Thompson, T. A. 2020, ApJ, 895, 43,doi: 10.3847/1538-4357/ab8ae8

Schneider, E. E., & Robertson, B. E. 2015, ApJS, 217, 24,doi: 10.1088/0067-0049/217/2/24

—. 2017, ApJ, 834, 144, doi: 10.3847/1538-4357/834/2/144Schweitzer, M., Lutz, D., Sturm, E., et al. 2006, ApJ, 649,79, doi: 10.1086/506510

Schweitzer, M., Groves, B., Netzer, H., et al. 2008, ApJ,679, 101, doi: 10.1086/587097

Shangguan, J., Ho, L. C., & Xie, Y. 2018, ApJ, 854, 158,doi: 10.3847/1538-4357/aaa9be

Shankar, F., Dai, X., & Sivakoff, G. R. 2008, ApJ, 687, 859,doi: 10.1086/591488

Silich, S., Tenorio-Tagle, G., & Muñoz-Tuñón, C. 2003,ApJ, 590, 791, doi: 10.1086/375133

Silva, C. V., Costantini, E., Giustini, M., et al. 2018,MNRAS, 480, 2334, doi: 10.1093/mnras/sty1938

Sim, S. A., Proga, D., Miller, L., Long, K. S., & Turner,T. J. 2010, MNRAS, 408, 1396,doi: 10.1111/j.1365-2966.2010.17215.x

Somalwar, J., Johnson, S. D., Stern, J., et al. 2020, ApJL,890, L28, doi: 10.3847/2041-8213/ab733d

Springob, C. M., Haynes, M. P., Giovanelli, R., & Kent,B. R. 2005, ApJS, 160, 149, doi: 10.1086/431550

Srianand, R. 2000, ApJ, 528, 617, doi: 10.1086/308214Srianand, R., Petitjean, P., Ledoux, C., & Hazard, C. 2002,MNRAS, 336, 753, doi: 10.1046/j.1365-8711.2002.05792.x

Stern, J., Faucher-Giguère, C.-A., Zakamska, N. L., &Hennawi, J. F. 2016, ApJ, 819, 130,doi: 10.3847/0004-637X/819/2/130

Stevans, M. L., Shull, J. M., Danforth, C. W., & Tilton,E. M. 2014, ApJ, 794, 75,doi: 10.1088/0004-637X/794/1/75

Sturm, E., González-Alfonso, E., Veilleux, S., et al. 2011,ApJL, 733, L16, doi: 10.1088/2041-8205/733/1/L16

Surace, J. A., Sanders, D. B., & Evans, A. S. 2001, AJ, 122,2791, doi: 10.1086/324462

Teng, S. H., & Veilleux, S. 2010, ApJ, 725, 1848,doi: 10.1088/0004-637X/725/2/1848

Teng, S. H., Veilleux, S., & Baker, A. J. 2013, ApJ, 765, 95,doi: 10.1088/0004-637X/765/2/95

Teng, S. H., Brandt, W. N., Harrison, F. A., et al. 2014,ApJ, 785, 19, doi: 10.1088/0004-637X/785/1/19

Thompson, T. A., Quataert, E., Zhang, D., & Weinberg,D. H. 2016, MNRAS, 455, 1830,doi: 10.1093/mnras/stv2428

Tombesi, F., Meléndez, M., Veilleux, S., et al. 2015, Nature,519, 436, doi: 10.1038/nature14261

Tombesi, F., Veilleux, S., Meléndez, M., et al. 2017, ApJ,850, 151, doi: 10.3847/1538-4357/aa9579

Tripp, T. M., Lu, L., & Savage, B. D. 1998, ApJ, 508, 200,doi: 10.1086/306397

Tripp, T. M., Sembach, K. R., Bowen, D. V., et al. 2008,ApJS, 177, 39, doi: 10.1086/587486

Trump, J. R., Hall, P. B., Reichard, T. A., et al. 2006,ApJS, 165, 1, doi: 10.1086/503834

U, V., Medling, A., Sanders, D., et al. 2013, ApJ, 775, 115,doi: 10.1088/0004-637X/775/2/115

Vanden Berk, D. E., Richards, G. T., Bauer, A., et al. 2001,AJ, 122, 549, doi: 10.1086/321167

Highly Ionized N V and O VI Outflows in the QUEST Quasars 37

Veilleux, S., Bolatto, A., Tombesi, F., et al. 2017, ApJ, 843,18, doi: 10.3847/1538-4357/aa767d

Veilleux, S., Kim, D. C., & Sanders, D. B. 2002, ApJS, 143,315, doi: 10.1086/343844

Veilleux, S., Maiolino, R., Bolatto, A. D., & Aalto, S. 2020,A&A Rv, 28, 2, doi: 10.1007/s00159-019-0121-9

Veilleux, S., Meléndez, M., Tripp, T. M., Hamann, F., &Rupke, D. S. N. 2016, ApJ, 825, 42,doi: 10.3847/0004-637X/825/1/42

Veilleux, S., Teng, S. H., Rupke, D. S. N., Maiolino, R., &Sturm, E. 2014, ApJ, 790, 116,doi: 10.1088/0004-637X/790/2/116

Veilleux, S., Kim, D. C., Peng, C. Y., et al. 2006, ApJ, 643,707, doi: 10.1086/503188

Veilleux, S., Rupke, D. S. N., Kim, D. C., et al. 2009a,ApJS, 182, 628, doi: 10.1088/0067-0049/182/2/628

Veilleux, S., Kim, D. C., Rupke, D. S. N., et al. 2009b, ApJ,701, 587, doi: 10.1088/0004-637X/701/1/587

Veilleux, S., Trippe, M., Hamann, F., et al. 2013a, ApJ,764, 15, doi: 10.1088/0004-637X/764/1/15

Veilleux, S., Meléndez, M., Sturm, E., et al. 2013b, ApJ,776, 27, doi: 10.1088/0004-637X/776/1/27

Vestergaard, M., & Peterson, B. M. 2006, ApJ, 641, 689,doi: 10.1086/500572

Virtanen, P., Gommers, R., Oliphant, T. E., et al. 2020,Nature Methods, 17, 261, doi: 10.1038/s41592-019-0686-2

Waddell, S. G. H., & Gallo, L. C. 2020, MNRAS, 498, 5207,doi: 10.1093/mnras/staa2783

Wang, T., & Wang, J. 1999, Publications of the PurpleMountain Observatory, 18, 248

Wang, T. G., Brinkmann, W., Wamsteker, W., Yuan, W.,& Wang, J. X. 1999, MNRAS, 307, 821,doi: 10.1046/j.1365-8711.1999.02650.x

Werk, J. K., Prochaska, J. X., Cantalupo, S., et al. 2016,ApJ, 833, 54, doi: 10.3847/1538-4357/833/1/54

Weymann, R. J., Carswell, R. F., & Smith, M. G. 1981,ARA&A, 19, 41,doi: 10.1146/annurev.aa.19.090181.000353

Weymann, R. J., Morris, S. L., Foltz, C. B., & Hewett,P. C. 1991, ApJ, 373, 23, doi: 10.1086/170020

Weymann, R. J., Turnshek, D. A., & Christiansen, W. A.1985, in Astrophysics of Active Galaxies andQuasi-Stellar Objects, ed. J. S. Miller, 333–365

Wills, B. J., Brandt, W. N., & Laor, A. 1999, ApJL, 520,L91, doi: 10.1086/312165

Wu, J., Brandt, W. N., Hall, P. B., et al. 2011, ApJ, 736,28, doi: 10.1088/0004-637X/736/1/28

Yi, W., Brandt, W. N., Hall, P. B., et al. 2019, ApJS, 242,28, doi: 10.3847/1538-4365/ab1f90

Zhang, S., Wang, H., Wang, T., et al. 2014, ApJ, 786, 42,doi: 10.1088/0004-637X/786/1/42

Zhuang, M.-Y., Ho, L. C., & Shangguan, J. 2018, ApJ, 862,118, doi: 10.3847/1538-4357/aacc2d

Zubovas, K., & King, A. 2012, ApJL, 745, L34,doi: 10.1088/2041-8205/745/2/L34

Zubovas, K., & King, A. R. 2014, MNRAS, 439, 400,doi: 10.1093/mnras/stt2472